University of Tennessee, KnoxvilleTrace: Tennessee Research and CreativeExchange

Doctoral Dissertations Graduate School

5-2010

Use of Proteomics Tools to Investigate ProteinExpression in Azospirillum brasilenseGurusahai K. Khalsa-MoyersUniversity of Tennessee - Knoxville, gsahaik@comcast.net

This Dissertation is brought to you for free and open access by the Graduate School at Trace: Tennessee Research and Creative Exchange. It has beenaccepted for inclusion in Doctoral Dissertations by an authorized administrator of Trace: Tennessee Research and Creative Exchange. For moreinformation, please contact trace@utk.edu.

Recommended CitationKhalsa-Moyers, Gurusahai K., "Use of Proteomics Tools to Investigate Protein Expression in Azospirillum brasilense. " PhD diss.,University of Tennessee, 2010.https://trace.tennessee.edu/utk_graddiss/712

To the Graduate Council:

I am submitting herewith a dissertation written by Gurusahai K. Khalsa-Moyers entitled "Use ofProteomics Tools to Investigate Protein Expression in Azospirillum brasilense." I have examined the finalelectronic copy of this dissertation for form and content and recommend that it be accepted in partialfulfillment of the requirements for the degree of Doctor of Philosophy, with a major in Life Sciences.

Gladys Alexandre, Major Professor

We have read this dissertation and recommend its acceptance:

Cynthia B. Peterson, Dale A. Pelletier, Jennifer L. Morell-Falvey

Accepted for the Council:Dixie L. Thompson

Vice Provost and Dean of the Graduate School

(Original signatures are on file with official student records.)

To the Graduate Council:

I am submitting herewith a dissertation written by Gurusahai Kaur Khalsa-Moyers

entitled “Use of Proteomics Tools to Investigate Protein Expression in Azospirillum

brasilense”. I have examined the final electronic copy of this dissertation for form and

content and recommended that it be accepted in partial fulfillment of the requirements for

the degree of Doctor of Philosophy, with a major in Life Sciences.

Gladys B. Alexandre___

Major Professor

We have read this dissertation

and recommended its acceptance:

Dale A. Pelletier___________

Cynthia B. Peterson_______

Jennifer L. Morell-Falvey_______

Accepted for the Council:

____Carolyn Hodges___

Vice Provost and

Dean of Graduate School

(Original signatures are on file with official student records.)

Use of Proteomics Tools to Investigate Protein Expression in

Azospirillum brasilense

A Dissertation Presented for

the Doctor of Philosophy Degree

The University of Tennessee, Knoxville

Gurusahai Kaur Khalsa-Moyers

May, 2010

ii

Dedication

To my daughter, Kelsey

For her never-ending encouragement and her happy smile.

iii

Acknowledgements

I would like to first acknowledge my advisor, Gladys Alexandre, for her time and

energy and for commenting on what seemed like endless versions of this work. I would

also like to acknowledge the members of my committee for their service: Dr. Dale A.

Pelletier, Dr. Cynthia Peterson, and Dr. Jennifer Morrell-Falvey.

I am grateful to a number of people for instruction and help with implementing all

techniques used in this work. Dr Dale Pelletier and members of his lab provided

laboratory facilities and instruction. Microbial growth assistance was provided by Tse-

Yuan Lu for R. palustris cultures, and DNA manipulation assistance was provided by

Linda Foote. Patricia Lankford helped tremendously with all protein preparation work

ranging from western blot procedures to trypsin digestion and all stages of mass

spectrometry. I am grateful for her ever-positive attitude and encouragement and

immense amount of help. I would also like to acknowledge Amber Bible from Gladys

Alexandre’s lab for help with growth of all Azospirillum cultures.

I would like to thank the members of the Organic and Biological Mass

Spectrometry group at Oak Ridge National Laboratory. I would like to specially thank

Keiji Asano, who was always available for help with any and all mass spectrometer

questions and issues. I am grateful to Greg Hurst for filling in when things were not

going so well, and for his help with mass spectrometry experiments, data analysis and

Microsoft Access. I would like to thank Miriam Land for help with BLAST searches.

I would like to acknowledge the support of the Genome science and technology

program at the University of Tennessee, Knoxville. And finally, this research was

iv

sponsored by the Genomics:GTL program, Office of Biological and Environmental

Research, U.S. Department of Energy, under contract No. DE-AC05-00OR22725 with

Oak Ridge National Laboratory managed and operated by UT-Battelle, LLC.

v

Abstract

Mass spectrometry based proteomics has emerged as a powerful methodology for

investigating protein expression. “Bottom up” techniques in which proteins are first

digested, and resulting peptides separated via multi-dimensional chromatography then

analyzed via mass spectrometry provide a wide depth of coverage of expressed

proteomes. This technique has been successfully and extensively used to survey protein

expression (expression proteomics) and also to investigate proteins and their associated

interacting partners in order to ascertain function of unknown proteins (functional

proteomics). Azospirillum brasilense is a free-living diazotrophic soil bacteria, with

world-wide significance as a plant-growth promoting bacteria. Living within the

rhizosphere of cereal grasses, its diverse metabolism is important for its survival in the

competitive rhizospheric environment. The recently sequenced genome of strain Sp245

provided a basis for the proteome studies accomplished in this work. After initial mass

spectrometer parameter optimization studies, the expressed proteomes of two strains of

Azospirillum brasilense, Sp7 and Sp245, grown under both nitrogen fixing and optimal

growth (non nitrogen fixing) conditions were analyzed using a bottom up proteomics

methodology. Further proteome studies were conducted with A. brasilense strain Sp7 in

order to ascertain the effect of one chemotaxis operon, termed Che1. In this study,

proteomic surveys were conducted on two bacterial derivative strains, created earlier,

which lacked either a forward signaling pathway or an adaptation pathway. The

proteomic surveys conducted in this work provide a foundation for further biochemical

investigations. In order to facilitate further investigation and a movement into functional

vi

proteomics, a set of destination vectors was created that contain a variety of tandem

affinity tags. The addition of tandem affinity tags to a protein allow for generic

purification schemes, and can facilitate future studies to investigate proteins of interest

discovered in the first expression proteomic surveys of A. brasilense. Taken together,

this dissertation provides a valuable data set for investigation into the physiology of A.

brasilense and further provides biochemical tools for analysis of the functional protein

interactions of A. brasilense cells.

vii

Table of Contents

Chapter 1. Introduction ..................................................................................................... 1

Proteomics tools......................................................................................................................... 1

Azospirillum brasilense ............................................................................................................ 20

Chapter 2. Optimization of Tandem Mass Spectrometry Methods ................................ 35

Introduction............................................................................................................................. 35

Materials and Methods........................................................................................................... 43

Results ...................................................................................................................................... 51

Discussion................................................................................................................................. 67

Conclusions.............................................................................................................................. 74

Chapter 3. Proteomic Investigations of A. brasilense strains Sp245 and Sp7 under

nitrogen fixing and non-nitrogen fixing conditions....................................................... 77

Introduction............................................................................................................................. 77

Materials and Methods........................................................................................................... 81

Results ...................................................................................................................................... 87

Discussion................................................................................................................................. 97

Conclusions............................................................................................................................ 135

Chapter 4. Comparison of Proteome Expression between Azospirillum brasilense wild-

type and mutants in a chemotaxis-like signal transduction pathway .......................... 138

Introduction........................................................................................................................... 138

Materials and methods ......................................................................................................... 139

Results .................................................................................................................................... 143

Discussion............................................................................................................................... 156

Conclusions............................................................................................................................ 180

Chapter 5. Development of a set of gateway-compatible destination vectors containing

C-terminal tags............................................................................................................... 182

Introduction........................................................................................................................... 182

Materials and Methods......................................................................................................... 185

Results .................................................................................................................................... 201

Discussion............................................................................................................................... 208

Conclusions............................................................................................................................ 221

Chapter 6. Conclusion .................................................................................................. 224

References ...................................................................................................................... 230

Vita.................................................................................................................................. 247

viii

List of Tables

Table 2. 1 Proteins included in the Extended Protein Standard Mix (EPSM).................. 44

Table 2. 2 Mass spectrometry operation parameters held constant .................................. 49

Table 2. 3 Variable tandem mass spectrometer experimental parameters........................ 49

Table 2. 4 numbers of protein and peptide identifications from total proteome spectra

collected using differing isolation widths ......................................................................... 57

Table 2. 5 Average numbers of protein and peptide identifications from total proteome

spectra collected using differing MS/MS microscan settings........................................... 62

Table 3. 1Total numbers of protein and peptide identifications in each sample .............. 88

Table 3. 2 Numbers and relative abundance percentages (by NSAF) of proteins identified

in each Sp245 sample by functional category................................................................... 90

Table 3. 3 Numbers and percentages of relative abundance of proteins (by NSAF)

identified in each Sp7 sample by functional category ...................................................... 92

Table 3. 4 Nitrogenase components and related proteins necessary for nitrogen fixation

detected in both nitrogen fixing and optimal growth control Sp245 cultures................... 95

Table 3. 5 Nitrogenase components and related proteins necessary for nitrogen fixation

detected in both nitrogen fixing and optimal growth control Sp7 cultures....................... 96

Table 3. 6 Fold change of proteins found to be up-regulated in Sp245 nitrogen fixing

cultures in comparison to optimal growth Sp245 controls ............................................... 98

Table 3. 7 Fold change of proteins identified in Sp245 nitrogen fixing cultures to be

down-regulated over Sp245 optimal growth control cultures......................................... 100

Table 3. 8 Fold change of proteins up-regulated in Sp7 nitrogen fixing cultures in

comparison to Sp7 optimal growth control cultures ....................................................... 102

Table 3. 9 Fold Change of Proteins down-regulated in Sp7 and Sp245 nitrogen fixing

cultures in comparison to optimal growth control cultures ............................................ 103

Table 3. 10 Fold Change of common proteins up-regulated in Sp7 and Sp245 molybdate

nitrogen fixing cultures in comparison to each respective optimal growth control culture

......................................................................................................................................... 104

Table 4. 1 Total number of protein identifications in each Sp 7 technical replicate run 145

Table 4. 2 Comparison of protein expression in the cultures of the parental strain Sp7 and

its ∆cheB1cheR1 and ∆cheY1 mutant derivatives by functional category..................... 146

Table 4. 3 Fold change of proteins up-regulated in both ∆cheB1cheR1 and ∆cheY1

mutants versus Sp7 controls............................................................................................ 149

Table 4. 4 Fold change of proteins down-regulated in both ∆cheB1cheR1 and ∆cheY1

versus Sp7 controls ......................................................................................................... 153

Table 4. 5 NSAF values of Type 6 Secretion System components detected in mutant

strains and in Sp7 control cultures .................................................................................. 155

Table 5. 1 C-terminal tags and sequences included in pBBR-based and pJQ200KS-based

vectors ............................................................................................................................. 189

Table 5. 2 Buffer composition for each affinity purification step ................................. 195

ix

Table 5. 3 Resins used for capture in each affinity purification step............................. 197

Table 5. 4 Sequence coverage of interacting partners for RpoA (rpa3226) detected using

mass spectrometry........................................................................................................... 205

Table 5. 5 Sequence coverage of interacting partners for RpoC (rpa3267) detected via

mass spectrometry........................................................................................................... 206

x

List of Figures

Figure 1. 1 Steps in proteomics work flows ....................................................................... 3

Figure 1. 2 Creation of fragment ions in tandem mass spectrometry A) CID energy

transfer process, B) Fragmentation patterns of peptides C) MS/MS spectra of peptide

VLDALDSIK from carbonic anhydrase. .......................................................................... 10

Figure 1. 3 Tandem affinity purification steps................................................................. 18

Figure 1. 4 Ribbon structure of nitrogenase enzyme complex from Azotobacter ........... 23

Figure 1. 5 Levels of regulation in expression of nitrogenase structural proteins and

related synthesis proteins .................................................................................................. 25

Figure 1. 6 Prototypical chemotaxis pathway in E. coli .................................................. 30

Figure 2. 1 Representation of spectral averaging (number of microscans) to obtain a final

observed spectrum ............................................................................................................ 40

Figure 2. 2 Representative fragmentation spectra from peptide sequence

DLILQGDATTGTNLELTR derived from protein conconavalinA collected using

different isolation widths .................................................................................................. 53

Figure 2. 3 Representative fragmentation spectra from peptide sequence derived from

protein concanavalin A collected using different microscans .......................................... 54

Figure 2. 4 ROC curves of LCQ isolation width EPSM data .......................................... 55

Figure 2. 5 ROC plots of LTQ isolation width EPSM data ............................................. 59

Figure 2. 6 ROC curves of pooled LCQ EPSM microscan data...................................... 60

Figure 2. 7 Box plot representation of statistical results from pooled LCQ proteome

microscan data .................................................................................................................. 63

Figure 2. 8 ROC curves representing LTQ pooled EPSM microscan data ..................... 65

Figure 2. 9 Box plots representing statistical analysis of LTQ microscan proteome data66

Figure 2. 10 Development of an ROC curve (Adapted from

http://www.anaesthetist.com/mnm /stats/roc/).................................................................. 69

Figure 3. 1 Proteome expression profile by relative abundance (NSAF) in Sp245 cells 91

Figure 3. 2 Sp7 protein expression by functional category in A) Sp7 optimal growth

controls and B) Sp7 nitrogen fixing cultures .................................................................... 93

Figure 3. 3 BLASTp alignment of alpha subunits of molybdate nitrogenase (AB1246)

and vanadium nitrogenase (AB6100) ............................................................................. 110

Figure 3. 4 Schematic representation of genomic region surrounding Abra_0139 and

Abra_0141....................................................................................................................... 116

Figure 3. 5 BLASTp alignment of protein sequences for hypothetical proteins abra_0139

and abra_0141................................................................................................................. 118

Figure 3. 6 Genomic region encompassing Abra_1839 MarR family transcriptional

regulator .......................................................................................................................... 125

Figure 4. 1 Proteome expression profiles for A) wild-type Sp7 control, B) mutants

lacking functional CheB1 and CheR1, and C) mutants lacking functional CheY1........ 148

Figure 4. 2 Pathway of PHB biosynthesis ..................................................................... 162

xi

Figure 4. 3 BLASTp depiction of conserved domains in abra_6288 ............................ 167

Figure 4. 4 Schematic depiction of genomic region containing type VI secretion system

homologs to IcmF (abra_0646) and and DotU (abra_0647)........................................... 175

Figure 4. 5 Schematic representation of extended genomic region surrounding the type

VI secretion system components in the Sp245 genome.................................................. 176

Figure 5. 1 Flow chart of vector construction steps....................................................... 190

Figure 5. 2 Schematic drawing of the initial pBBR-Dest42 parent vector .................... 202

Figure 5. 3 New set of pBBR-Dest42-based vectors showing expansions of the three new

C-terminal tag constructs ................................................................................................ 203

Figure 5. 4 Schematic depiction of RNA polymerase subunit interacting partners

(http://string.embl.de) ..................................................................................................... 222

xii

List of Attachments

Sp245N2Fix_Data

Sp7N2Fix_Data

2Fold_Down_N2Fix

2Fold_Up_N2Fix

Mutant_Proteome_Data

1

Chapter 1. Introduction

Proteomics tools

Biochemical techniques provide a wealth of information regarding the physiology

of individual cells in terms of individual proteins and their structure, function and

location within a cell, and participation within protein complexes and functional

pathways. Combining this knowledge with systems biology information can serve to

broaden the perspective of cell physiology under specific conditions. Systems biology

approaches include acquisition of genome structure (genomics), which gives a picture of

what is possible for a cell, as well as elucidation of which genes are transcribed in a given

situation (transcriptomics). Further, systems biology approaches include a number of

techniques that investigate the average population of proteins present within a cell as well

as their modifications and possible function. The “expression proteome” of a cell

consists of the entire set of proteins expressed in cell, while the “functional proteome”

examines the protein-protein interactions on a genome-wide scale in an attempt to dissect

functional pathways [1]. From these definitions, it follows that proteomics is defined as

the study and identification of the full complement of proteins expressed in a cell under a

given set of conditions. Due to its sensitivity and its ability to identify a large percentage

of the proteins present in complex mixtures, mass spectrometry has emerged as a very

powerful experimental tool for investigation of proteomes.

Shotgun proteomics is a term derived from DNA shotgun sequencing [2] and

represents a methodology in which proteins in a mixture are digested with proteases to

yield a very complex peptide mixture. Peptides within this mixture are then separated

2

and analyzed using mass spectrometry, allowing corresponding proteins from which they

were derived to be identified following analysis [3-5]. In tandem mass spectrometry

experiments, parent peptide ions are first isolated and then fragmented through collision

with an inert gas, giving rise to raw fragmentation spectra containing unique information

about the parent peptide sequence [3]. Using database searching algorithms, individual

peptide sequences can be derived from comparison of the fragmentation pattern of parent

peptide ions with theoretical spectra derived from an available database [3, 6, 7]. When

sequence information is not available, other algorithms deduce peptide sequences or parts

of peptide sequences from the spectral information and then compare these “tags” to

protein sequences derived from larger databases composed of sequence information from

multiple closely related organisms to make protein identifications [8]. The primary goal

of this dissertation was to employ a mass spectrometry-based proteomics work flow using

shotgun sequencing methods to explore the expression proteome of Azospirillum

brasilense cells.

Proteomics work flows

For most biologists, mass spectrometry represents a “black box” from which large

amounts of data emerge. Within this black box are contained a number of different steps

which make up an entire proteomics work flow as represented in Figure 1.1. At each step

of the proteomic work flow, decisions need to be made and optimizations done in order to

ensure the most useful and instructive data output. First and foremost, if investigating

proteomes from bacterial isolates or eukaryotic cell lines, decisions about growth need to

be made. First, the size of the culture needed for total proteome investigation or for

3

Figure 1. 1 Steps in proteomics work flows

A number of individual steps are involved in a proteomics work flow. Decisions and optimizations at each

step of the work flow are made based on the questions being asked. Amount of material needed for

individual experiments dictates the culture size and conditions. Mass spectrometry platforms chosen

include such decisions as separation techniques (online or off-line), ionization techniques (typically

MALDI or ESI), and type of mass spectrometer used for analysis. Once data is obtained, choices must then

be made about search algorithms used and types of statistical analysis applied. From this analysis, filtering

levels must be chosen that are applicable to each specific data set to ensure significance of results.

Ionization

Off-line, Online, Single phase, Multidimensional (MudPIT)

Sample Preparation

Mass Analysis

Data Analysis

Separation

Lysis

Growth

Small Sample Preparation, Lysis Buffers

Isolate vs Environmental sample, Amount, Conditions, Replicates, Timing, Metabolic labeling

Immunoprecipitation for protein complex investigation, Affinity Isolation, PAGE, 2D-GE, Protease digestion, Cleanup

Ion Traps (Quadrupole, Linear Quadrupole) Time-of-Flight (TOF)

MALDI ESI, nano-ESI

De novo sequencing Database search Filter and sort results

4

isolation of protein complexes needs to be determined. For example, nitrogen fixing

cultures had to be grown in a relatively small volume in order to minimize exposure to

oxygen, which is detrimental to nitrogenase function. Culture volumes were therefore

limited to 250 ml media in a 1liter flask. Azospirillum brasilense nitrogen fixing cells

will grow only to an optical density of 0.1-0.2 under these conditions, yielding a wet cell

pellet weight of ~0.05g. Resolubilization of this pellet in 250 µl buffer yields ~1 mg/ml

protein concentration, which after clean-up and exchange into mass spectrometry

compatible buffers will yield enough protein lysate at a concentration of ~ 1mg/ml for

only 2-3 technical replicates. As a result, 2 or more biological replicates must be

combined in order to have enough protein for multiple technical replicate mass

spectrometry runs. More typically, larger cultures of bacterial isolates ranging from 500

ml to 1.5 L are grown for expression proteomes. A 500 ml A. brasilense non nitrogen

fixing culture grown to mid-log phase (OD600 ~ 0.6) yielded a wet cell pellet weight of

~0.3 g. After resolubilization in 1.5 mL buffer, a protein concentration of ~1mg/ml was

obtained and enough lysate was obtained for 6-7 technical replicates.

Additional decisions about growth include the culture media to be used for

growth, the possible treatments or growth conditions to be applied to each culture if

doing proteome comparisons, the number of replicates needed to establish statistical

validity, and the amount of time required for growth, especially if doing a time course

study. Optimization of each of those parameters is unique to the organism under study

and the research question being addressed which will ultimately dictate the set of

conditions and the type of experiments being performed. An additional choice at this

initial step in the process can involve addition of labeling agents such as “heavy”

5

nitrogen, 15

N [9], or “heavy” labeled amino acids [10], to the culture medium in order to

shift the mass of proteins within a given sample. This labeled sample can then be mixed

with a control sample during sample preparation steps, and the resulting mix used for

direct comparisons of relative amounts of individual proteins per growth state or for

ascertaining non-specific interactions between proteins [11, 12].

The choice of growth conditions combined with the desired downstream analysis

dictates the lysis method used to retrieve the proteins from the cell for further analysis.

For example, small sample sizes benefit from single tube lysis procedures such as those

developed by Thompson et al [13]. Buffer compositions for lysis buffers need to be

chosen carefully to facilitate downstream processing, as illustrated by the need for

specialized or optimized lysis buffers to accommodate downstream processing in

immunoprecipitation experiments or affinity isolations of single proteins. For instance,

detergents found within some chemical lysis buffers can interfere with the ability of

tagged proteins to bind to affinity resins (Qiagen, Valencia CA). Protein concentrations

of samples need to be optimized through choice of lysis methodology and subsequent

downstream concentration and clean-up steps, since both steps can lead to sample loss.

Although mass spectrometry can detect proteins present at low femtomole levels within a

sample [14], reproducibility of spectra acquired and ability to reliably detect and identify

all peptides and thus proteins present suffers when the concentrations of proteins are low

[15].

Further sample preparation methods may involve a number of steps. Complex

protein mixtures separated in chromatographic steps in which peptides are eluted directly

6

into the mass spectrometer (online chromatography) by methods such as

Multidimensional Protein Identification Technology (MudPIT) [16-20] used in Chapters

2-4 of this work, may require only digestion and sample clean-up as further sample

preparation steps. Since online separation methods minimize sample loss and can also

minimize contamination from outside sources [21], this is an excellent methodology to

use for small sample sizes. Liquid chromatographic separations coupled directly to the

mass spectrometer can include both single dimensional separation, usually reverse phase,

and multiple dimension separation [22]. For simple samples derived from a single

purified protein or protein complex, a single dimension of separation may yield enough

depth of coverage of the available proteins. However, if a more complex sample is being

investigated, such as the subset of membrane proteins [22-24] or the total soluble

proteome [19, 25-33], then one or more separation steps are demanded as part of the

sample preparation in order to reduce sample complexity and improve the depth of

proteome coverage. Reproducibility of chromatographic separation is of great interest,

with a number of studies investigating the degree of reproducibility of both

chromatographic retention times and subsequent mass spectra derived at a specific

retention time in replicate runs [34, 35]. This may be especially significant when

comparing two different growth states because retention times of the same peptide

identified in both mass spectrometric runs can be used as validation of data results [35].

Alternatively, when a sample has a great deal of complexity, off-line separations

in which chromatography is performed and fractions collected for later mass

spectrometry analysis can also be performed. A single dimension separation of proteins

via SDS-PAGE gel electrophoresis or multi-dimensional separation via 2-dimensional gel

7

electrophoresis (2D-GE) are common steps in sample preparation [36]. This type of

separation allows the proteins present to be directly visualized which may also provide a

visual quantification of change in protein expression levels. At the same time, gel

separation methods are limited by the protein concentrations present within the sample

because of limits of detection of protein stains used in gels. Inherent physical properties

of proteins may also prove to be a limiting factor in this separation methodology since gel

electrophoresis tends to be biased for proteins within a specific pI range or molecular

weight range [36]. Once visualized, the protein bands can be excised and in-gel digests

performed before extracting peptides for subsequent mass spectrometry analysis. Due to

the hydrophobic nature of membrane proteins and difficulties with solubility in liquid

separations, separation by gel electrophoresis before analysis is the preferred

methodology for initial separation of membrane proteins [23]. However, from personal

experience, recovery of digested peptides from gels is poor. As a result, optimization of

growth and lysis procedures, as well as subsequent preparation steps for maximal protein

expression, is important for sufficient data output [37]. Off-line protein separation can

also be accomplished through High Performance Liquid Chromatography (HPLC)

separations with subsequent examination of the proteins present in individual fractions by

mass spectrometry. This is a commonly used methodology for examination of intact

proteins [38], or investigation of modifications such as phosphorylation [39] because it

can allow enrichment of the desired protein or of specific protein modifications in a

mixture, thus optimizing its concentration and subsequent mass spectrometry.

Once the samples are prepared, the “black box” of mass spectrometry, including

ionization techniques, mass analysis and mass detection, can be applied. A number of

8

excellent reviews [4, 5, 40-43] cover each of these elements of mass spectrometry, so

they will be only briefly mentioned here. Investigation of large biomolecules was

facilitated by the development of “soft ionization techniques” of Matrix Assisted Laser

Desorption Ionization (MALDI) [44] and electrospray ionization (ESI) [45], which

allowed for transfer of biomolecules into the gas phase without fragmentation. These two

ionization techniques are fundamentally different. When using a MALDI source,

proteins or peptides are first mixed with a matrix material, spotted onto a plate, and dried.

The plate is placed onto the MALDI source in a vacuum and a laser pulse is used to

ablate both matrix and analyte molecules from this solid surface to form ions, which then

enter the mass analyzer. Electrospray ionization is accomplished through application of

voltage to a needle at atmospheric pressure through which preformed ions in a liquid

phase are transformed into the gas phase and subsequently sprayed into the mass

analyzer.

The most commonly used mass analyzers in proteomics experiments are Time-of-

Flight (TOF) and ion trap mass analyzers [5]. MALDI ionization sources are most often

coupled with TOF analyzers, which measure the mass-to-charge ratio (m/z) of the ions

present by detecting the time it takes an ion to traverse the length of a flight tube under a

vacuum [5]. A high voltage is applied to the MALDI source and ions enter the flight tube

and travel down the length of the tube to the detector, with larger molecules traveling at

slower rates than smaller ones. Mass is determined by the length of time it takes the ion

to traverse the flight tube. ESI, and its lower flow-rate sister, nano-ESI, are most often

coupled with ion trapping mass analyzers, either linear ion trapping quadrupoles or three-

dimensional trapping quadrupoles [5]. Quadrupole instruments trap ions in either 2 or 3

9

dimensions through the application of radio frequency (rf) and DC voltages. Detection of

ions stored within the trap is accomplished by ramping voltages such that ions of

decreasing mass-to-charge ratio become unstable and exit the trap where they are then

detected as they hit an electron multiplier tube. See March [46, 47] and Schwartz [48] for

thorough reviews of ion trap operation and theory. Although there are a number of other

mass analyzers, including hybrid configurations of the above-mentioned mass analyzers,

they will not be mentioned here.

Tandem mass spectrometry (MS/MS) is a powerful technique for investigating

protein sequence. Ion trapping instruments lend themselves well to tandem mass

spectrometry experiments because of ease in trapping and isolating ions of interest

although a number of other mass analyzers can also perform MS/MS. During tandem

mass spectrometry experiments in an ion trap, the parent ion of interest is isolated within

the trap and then fragmented, often through a process known as collision-induced

dissociation (CID) [49]. In a CID experiment (Figure 1.2), the isolated parent ions are

subjected to collision with a neutral inert gas. Energy from repeated collisions is

transferred to the ion causing it to vibrate. When enough energy is accumulated, the

peptide backbone bonds break, primarily at amide bonds resulting in b and y type ions

(Figure 1.2). The resulting fragment ions are then measured and compiled to give a

series of peaks representing the measured mass-to-charge (m/z) ratio of the product ions

[49]. The distance between the individual peaks in a spectrum represents the mass of an

amino acid, thus giving rise to sequence information, as represented in Figure 1.2.

10

Figure 1. 2 Creation of fragment ions in tandem mass spectrometry A) CID energy transfer process,

B) Fragmentation patterns of peptides C) MS/MS spectra of peptide VLDALDSIK from carbonic

anhydrase.

A) Analyte molecules, represented by large filled circles, collide with an inert gas, represented by small

open circles, causing a buildup of energy in the molecule that causes rupture of the weakest bonds of the

analyte molecule. B) Fragmentation patterns of peptide molecules are predicted by common breakage

points. In CID experiments, the peptide ion breaks most commonly at the peptide bond, creating “y” and

“b” type ions, represented above by the dotted lines above which are y1, y2, or y3 and below which are b3,

b2, or b1. The direction of the slash above the dotted line indicates which ion fragment retains the charge,

thus making it “visible” to the mass spectrometer. Thus y-type ions retain charge on the C-terminal end and

b-type ions on the N-terminal end. C) Sequence determination is accomplished through examination of the

distance by which individual peaks are separated from one another. The above spectrum shows“y-type”

ions created through a CID process. The first visible peak in the spectrum at 347.25 m/z is the y3 ion made

up of amino acids lysine, isoleucine and serine. The next peak used to identify the peptide sequence is at

462.28 m/z (labeled above as y4). The fragment ion contributing this peak is made of amino acids K, I, S,

and aspartic acid, D. The mass of aspartic acid is 115.09 Da (shown above the D), and thus the y3 and y4

peaks are separated by 115 Da due to the addition of the D amino acid in the fragment ion sequence. The

y5 ion at 575.34 m/z is composed of KISD and leucine (L). The molecular weight of leucine is 113.16 Da

and thus the peaks of the y4 and y5 fragment ions are separated by 113 Da. Fragment ion peaks for each

additional peak labeled in the spectrum above are separated by the individual weights of the additional

amino acids. When search algorithms are assigning peaks, the experimental spectra such as that shown

above are compared to theoretical spectra containing peptide fragment ions derived from sequences within

the database. The distance of the peaks from one another allows the search algorithm to assign sequence

information to the experimental spectra complete with a score that represents how well the experimental

spectrum matches the theoretically derived spectrum.

A) Collision induced dissociation B) Peptide Fragmentation Pattern

C) Sequence determination from a tandem mass spectrum

V L D A L D S I K

y3 y4 y7 y6 y5 347.43 462.52 575.68 646.76 761.85

D L KIS D A

N-terminal C-terminal y3 y4 y5 y7 y6

115.09Da 115.09Da 113.16Da 71.08Da

11

Once raw MS data has been acquired, interpretation of individual spectra can be

done manually or by automated processing strategies. A number of software programs

have been developed to extract information from raw mass spectra, thus facilitating fairly

rapid analysis of MS and MS/MS results. For those organisms which have no genome

sequence available, de novo programs such as GutenTag [8], DirecTag [50] or Lutefisk

[51] derive sequence information directly from the raw spectral data, with the distance

between peaks in MS/MS spectra representing the mass of individual amino acids in a

sequence. Sequence tags consisting of 3 or more amino acids identified by distances

between abundant peaks in the experimental spectrum can then be compared to a large

database, such as the NCBI non-redundant database [52], to make possible protein

predictions. When genomic sequence information for an organism is available, protein

databases can be developed based on open reading frames which predict proteins from

genomic sequence. Database search algorithms such as SEQUEST [3] or MASCOT [53]

can then be used to compare raw spectra to theoretical spectra derived from in silico

digest of the protein sequences within the database, with scores given for each available

match. Programs such as DTASelect [54] then filter and sort the data output from the

search algorithm, putting final protein lists together in an easily readable format.

Regardless of the type of experiment done, the quality of raw data collected will

certainly affect the ability of the search algorithms to make true positive identifications.

When surveying an entire proteome of a cell, a vast amount of data is acquired and the

possibility for false positive identifications is high. Studies have been done both to

address the issues of optimization of data quality being input to search algorithms [15, 34,

12

55-61] and of optimization of filter levels to ensure quality data output after searches are

performed [62].

In a tandem mass spectrometry experiment, a number of parameters can be

manipulated by the experimenter to ensure an optimal data output. Since ion trap

instruments were used in the remainder of this work, we will only consider parameters

manipulated in these instruments. In an ion trap, these parameters include amount of

time required to fill the ion trap [57, 60], scan rates [34], collision energy required for

fragmentation of the parent ion [57], parent ion isolation m/z window, and number of

averaged scans (microscans) contributing to the final scan [58-60]. The ion trap fill time,

scan rates, collision energy, and number of averaged scans have been investigated, with

optimized ranges of operation being determined. The experiments investigating the

optimum number of averaged scans [58-60] have been controversial, and the width of the

isolation m/z window chosen for isolation of a parent ion has not been investigated at all.

In Chapter 2 of this thesis, two tandem mass spectrometry parameters, number of

microscans and width of the isolation window selected when isolating the parent ion, will

be discussed in more detail. The isolation m/z window width and the number of averaged

scans (microscans) were optimized for experiments done on both a linear ion trapping

quadrupole instrument (LTQ) and a 3-dimensional trapping quadrupole instrument

(LCQ). We began our optimization with a mixture of known proteins in which we could

get definitive identifications, and then tested our parameters using a total proteome

preparation of the soil bacterium, Rhodopsuedomonas palustris. Results from this work

13

were then applied to subsequent investigation of the soluble proteome of two different

strains of Azospirillum brasilense, described more thoroughly below.

Mass spectrometry based proteomics in bacterial systems

The last few years have seen an explosion of mass spectrometry-based proteomics

experiments. A simple search in PubMed using the words “microbial proteomics”

returns 418 articles, with over 80 of those articles being published in the last year. The

goal of many earlier proteomics experiments in bacteria and archaea was to simply

ascertain the protein complement of growing cells, often with the added goal of

contributing to the genome annotation of newly sequenced strains [25, 26, 63, 64]. More

recent studies have not only characterized the complement of proteins expressed within

growing cells, but also have included comparisons of the cells grown under different

conditions, such as comparing the proteome expression profiles of R. palustris cells

grown under different metabolic states [30] or the comparison of the proteomes of Nostoc

punctiforme cells grown under nitrogen and non nitrogen fixing conditions [29]. Other

studies seek to determine proteins involved in pathways contributing to changes in

bacterial function by detecting changes in expression levels of proteins when microbes

are exposed to different growth environments, such as identifying a putative set of

proteins involved in nodulation [26] or pathogenicity [65] or metabolism of alcohol [66].

There are a number of ways to perform comparative analysis of proteins within

samples, either using labeling techniques or using label-free techniques. Labeling of

samples for subsequent mass spectrometric analysis allows for direct comparison of

individual protein abundance between two different growth states by comparison of the

14

mass shift induced in labeled peptides with their unlabeled counterpart. Labeling can be

accomplished at different points in the sample preparation pathway. For instance, mass

shifts can be introduced into proteins by introducing a heavy isotope during growth such

that all proteins synthesized during growth will incorporate the heavy isotope [9, 10].

Alternatively, proteins can be labeled after harvest by addition of labeling agents such as

heavy isotope 18

O to samples. In this instance, proteins are proteolytically digested, dried

and resolubilized in “heavy” water containing a heavy isotope of oxygen 18

O. Two

molecules of heavy oxygen are incorporated into the carboxy groups at the ends of the

peptide, thus shifting the peptide mass by 4 Da [67]. Also late in the preparation

pipeline, at the final step in preparation, peptides can be labeled with the isobaric reagent,

iTRAQ®, which does not cause a mass shift, but instead releases a signature reporter ion

on fragmentation, allowing for quantification of abundance in the labeled sample [68].

Recently, this methodology was used to quantify changes in 721 proteins in Nostoc

punctiforme, both filament and heterocyst, upon shifting to nitrogen fixation [29]. The

changes in individual protein abundances were detected via presence of reporter ion in

the MS/MS scans. The proteins showing changes in abundance were mapped back to

metabolic pathways, and allowed confirmation of existing pathway data as well as

discovery of novel participants in metabolic pathways affected by a shift to nitrogen

fixation. Although this study is an excellent example of the use of iTRAQ technique, our

lab has found the reagents expensive and difficult to use, with the further complication of

inefficient labeling of proteins.

An interesting example of a labeling technique for bacterial cultures termed I-

DIRT [9] includes growth of a test culture in media containing heavy isotope components

15

of 15

N nitrogen alongside a control culture that does not contain the heavy isotope. The

two cultures are harvested and mixed together in equal weights, and the mixed cultures

are then analyzed using mass spectrometry. Use of heavy nitrogen allows all proteins

within the test sample to be labeled in such a way as to cause a shift in mass in the test

culture. When mixed with a non-labeled culture and analyzed via mass spectrometry,

twin peaks will be observed for peptides present in both samples, one from the non-

labeled control, and a second mass-shifted peak from the labeled test sample. Calculation

of the peak area for each of these spectra gives the measure of relative abundance of the

peptide within the heavy labeled test sample in direct comparison to relative abundance

within the non-labeled control. Recently this labeling technique was applied to a global

analysis of E. coli and R. palustris cultures with protein detection done using MudPIT

[11]. In addition to providing information about individual proteins and their interacting

partners after immunoprecipitation experiments, quantitation of all proteins detected was

done to assess the global effect of expressing tagged proteins off a plasmid. Direct

comparison of individual peptide areas can be done to provide an accurate comparison of

relative abundance in different cultures.

However, in spite of the success of labeling techniques in quantification of

relative abundances of proteins between two samples, one labeled and one unlabeled, the

labeling reagents are expensive, and data analysis of mixed samples is often time

consuming and quite complicated [69]. Label-free techniques have the advantage of two

cultures being grown in media with similar components, such that any possible

interference from heavy isotopes during growth is avoided. In addition, label-free

approaches can offer the advantage of relatively uncomplicated data analysis. In shotgun

16

proteomics experiments analyzed by using MudPIT [20], spectral count of peptides has

been shown to correlate well with relative abundance of a protein present within an

individual sample [70]. Spectral abundance factors (SAF) can be calculated by taking

into account how often a given peptide is sampled by the mass spectrometer (spectral

count or SpC) divided by the total length of the protein. Dividing by the length of the

protein accounts for the probability of higher spectral counts due to a higher number of

peptides present with increasing length of the protein [71]. Since different replicates will

have different numbers of identified peptides and proteins, normalization of SAF (NSAF)

for each individual run allows for comparison of relative protein abundance across

samples [72]. In Chapter 3 of this work we have used label-free shotgun proteomics to

investigate the baseline proteomes of A. brasilense, a newly sequenced organism.

Further, NSAF values have been calculated and used to investigate the expression levels

of proteins detected when isolates were grown under both nitrogen fixing conditions and

non-nitrogen fixing conditions (see below). While chapter 3 investigates the effect of

differing environmental conditions, chapter 4 of this work utilizes the same proteomic

survey method to examine the effect of mutations in one specific chemotaxis-like

pathway (see below) on the overall proteome expression in A. brasilense Sp7 cells.

Functional proteomics

By nature, expression proteomics is a discovery-based technique, simply

elucidating the proteins that are present at a given time under specific growth conditions.

This survey can in turn produce insights into physiological function and can lead to

identification of target proteins that need further functional exploration. One

methodology for exploration of protein function includes examination of the interacting

17

partners of a particular protein. Proteins often function within a cell as part of a

“molecular machine” with associated proteins all serving to contribute to the overall

outcome of the “machine”. One such example includes DNA-directed RNA polymerase,

in which 5 subunits associate and function together (along with a number of associated

proteins) to transcribe DNA into messenger RNA. Thus the function of individual

proteins can often be elucidated by determining the molecular machinery (protein

complex) with which it participates [1]. Classic gold-standard biochemical techniques

for determining protein interacting partners on a large scale include yeast 2 hybrid

techniques and phage display techniques [1]. These, however, are limited to direct

pairwise interactions and investigations are often dictated by what is currently known.

Mass spectrometry-based techniques, on the other hand, are not limited to detection of

only pairwise interactions, but instead can detect all proteins present within a complex

[16]. Recently developed techniques include strategies for tagging proteins with an

affinity epitope such that proteins and their associated interacting partners can be isolated

on a genome-wide scale, thus facilitating development of large protein interaction

networks for model organisms such as E. coli [73-75] or S. cerevisiae [16, 76-80], as well

as in soil bacterium R. palustris [81]. A strategy in which a protein of interest in fused to

an affinity purification tag provides an efficient means with which to isolate a protein and

associated interacting partners [82]. Since single affinity tags can lead to isolation of a

large number of non-specifically interacting proteins, tandem affinity tagging systems

have been developed in which two affinity tags are fused to either the C-terminal or N-

terminal end of a protein, thus allowing for purification of proteins to near homogeneity

18

Figure 1. 3 Tandem affinity purification steps

Proteins tagged with a tandem affinity tag can be purified in a two step-purification process. Cell lysate

containing a tagged bait protein is affinity purified using first via the outer tag. In this purification step, the

bait protein is tightly bound to the resin, and a number of non-specific interacting proteins are present in the

mixture bound to both the bait protein and to the affinity resin. The outer tag is then cleaved off, leaving

the bait protein and its associated interacting partners free in solution. The second affinity purification is

done using the remaining inner affinity tag. After incubation, washes help to remove a majority of non-

specifically bound proteins. Gentle elution and subsequent proteolytic digestion then prepare the proteins

and interacting partners for detection via mass spectrometry. All steps are performed using physiological

buffers.

Protein complex

in cell lysate

1st affinity purification- outer tag

2nd affinity purification- inner

tag

Protease cleavage of outer tag-

bait in solution

Elution of bait protein

Digestion

MS

detection

19

and decreasing the numbers of non-specifically interacting proteins [75, 76, 80, 83, 84].

Inclusion of a protease cleavage site between the tags allows for purification strategies

done under physiological conditions, as represented in Figure 1.3. When fused to a

tandem affinity tag containing a protease cleavage site between the two motifs, the

protein of interest (“bait” protein) is first purified by immunoprecipitation using the

outermost high affinity tag. For this initial purification, physiological buffers are used,

thus preserving non-covalent interactions between protein interacting partners. The

innermost tag is then cleaved off through incubation with an appropriate protease, again

in physiological buffers, leaving the tagged protein and its associated interacting partners

free in solution. A second immunoaffinity step is then performed, concentrating the

protein and its associated interacting partners and further washing non-specifically

interacting proteins away. The innermost affinity tag is chosen such that gentle elution

conditions may be employed from this affinity purification step. The use of physiological

buffers in all steps of the purification strategy, combined with the gentle elution

conditions, serve to facilitate preservation of non-covalent interactions between the

protein of interest and its interacting partners, thus preserving protein complexes that may

be lost when harsher purification and elution conditions are employed [1]. Disadvantages

to affinity tags include possible interference of the tag with protein folding or structure,

or interference of the tag with interaction sites [85]. Further, since proteins contain a

wide variety of sequences and structures, expression levels can differ with different tag

moieties or a protein can be insoluble with one tag but not another [85]. In order to

increase functional proteome coverage, it is therefore desirable to have a set of tags with

which to tag proteins. With that in mind, in Chapter 5 of this work, we present

20

development of a set of destination vectors based on pBBR-Dest42 created earlier [81].

The advantages of this vector set include the broad host range expression afforded by the

pBBR replicon of the parent vector and ease of cloning through the inclusion of a

Gateway cloning cassette in the parent vector. The new vector set includes a choice of

C-terminal tags based on tandem affinity tags that have been previously proven to

function efficiently to isolate protein complexes in bacterial and eukaryotic systems.

RNA polymerase subunits were then expressed as a fusion with each individual tag and

used as bait proteins in immunoprecipitation experiments with generic protocols. Prey

proteins that co-immunoprecipitated with the bait were then detected via mass

spectrometry as a proof of concept for the vector set. Results are presented in Chapter 5.

Azospirillum brasilense

Rhizospheric habitats are defined by the soil regions surrounding plants in which

are found abundant root secretions in a concentration gradient that varies inversely with

increasing distance from the plant [86]. This habitat offers a great variety of nutrients

that serve as a source of both carbon and energy for bacterial cells that live within the

rhizosphere. Plant root exudates include organic acids, amino acids, sugars, vitamins,

and enzymes [86]. The specific composition of the nutrients found within a given soil

environment is dependent upon the plant growing therein. Further, this composition

changes as a function of distance from the root [86]. Motile bacteria such as Azospirillum

species with ability to utilize a number of different carbon and energy sources possess a

competitive advantage in these environments. Interaction with the plant root hair is

dependent upon the ability of the bacteria to show biased movement, or chemotaxis, to

21

the root exudates [87]. Additionally, the bacterial surface composition facilitates

interaction with the root hairs and further mediates formation of bacterial aggregates

which can facilitate nitrogen fixation capabilities [88]. The ability to fix nitrogen and to

show chemotaxis to optimal environments are important adaptation strategies to the

challenge of living in the rhizosphere where competition for limited nutrients is fierce.

Azospirillum brasilense are free-living soil bacteria that are found within the

rhizosphere of grasses and cereals in tropical and sub-tropical climates. They are a

member of the alpha subclass of proteobacteria, and are capable of promoting plant

growth upon inoculation of the roots of cereal grasses and grains [89]. Azospirillum

colonizes the root hair zone where easily metabolizable nutrients such as organic acids

and low molecular weight sugars are abundant [86]. Azospirillum species can use a

number of carbon and nitrogen sources. They are motile, and show taxis to root

exudates, but with primary taxis response to optimal low oxygen levels [90], allowing

them to adapt well and outcompete other bacterial species found within the rhizosphere.

Further, although they do not survive well in nutrient poor environments, they are able to

store carbon for later use in the form of poly-3-hydroxybutyrate (PHB) granules and also

to form cyst-like structures that facilitate survival under less than optimal conditions [86].

Although they do fix nitrogen, they do not release large amounts of ammonia into the

soil, so their plant growth promotion is thought to be primarily due to the secretion of

plant-growth promoting hormones such as indole-3-acetic acid (IAA) that increase

volume and number of root hairs, thus leading to better uptake of nutrients and creating a

stronger, healthier, bigger plant [91]. Because of their physiological adaptability and

22

positive effects on growth of cereal grains, Azospirillum species are of considerable

interest as inocula for cereal crops world-wide [91, 92].

In chapter 3 of this work, the total proteomes of two different strains of A.

brasilense, Sp245 and Sp7, grown under both nitrogen fixing and non nitrogen fixing

conditions are investigated. A. brasilense strain Sp245 was originally isolated from

wheat roots [89] and has been shown to colonize both the outer surface as well as the

interior of wheat root hairs [93]. A. brasilense strain Sp7 was originally isolated from the

rhizosphere of Digitaria decumbens (Brazil) [94], and colonizes only the outer surface of

root hairs [93]. Since each strain occupies a different ecological niche, it is likely that

each will respond differently to environmental challenges. Exploring the complement of

expressed proteins within each strain under both nitrogen fixing and non nitrogen fixing

conditions will contribute to understanding the differences in metabolic profiles and

physiology of these strains.

Nitrogen fixation

Although air contains almost 80% nitrogen which diffuses into the soil, it is not in

a form that is readily available for plant uptake. Therefore, nitrogen-fixing soil bacteria,

either free-living or endophytic, are essential for converting nitrogen gas to a more

readily available form. This is accomplished through the activity of the prokaryotic-

specific nitrogenase enzyme complex encoded by the nif operon, which encodes

dinitrogenase reductase (also known as Fe protein) and dinitrogenase (also known as

FeMoCo or FeMo protein) [91]. Dinitrogenase, a heterotetramer of two subunits

containing a molybdenum-iron (FeMo) cofactor (Figure 1.4), is responsible for reduction

23

A) PDB 1M34

B. PDB 1M1Y

Figure 1. 4 Ribbon structure of nitrogenase enzyme complex from Azotobacter

A) Protein data bank (www.rscb.org) accession number 1M34 Nitrogenase complex from

Azotobacter Vinelandii stabilized by ADP-tetrafluoroaluminate B) Protein data bank

accession number 1M1Y Chemical crosslink of nitrogenase MoFe protein and Fe protein

24

of nitrogen to ammonia, which is then assimilated into the amino acids and proteins

within the bacterial cell through the action of glutamine synthetase. Dinitrogenase

reductase, a homodimer containing an iron (Fe) cofactor (Figure 1.4), serves to donate

electrons from ferredoxin to dinitrogenase [91]. This energetically expensive process is

tightly regulated on several levels from translation of nitrogenase structural components

to post-translational modification of the nitrogenase enzyme complex [95, 96]. The

nitrogen regulation system (Ntr operon) regulates nitrogenase expression based on

general nitrogen metabolism within the cell. There are four proteins within the Ntr

system system: dual function uridylyltransferase/uridyl-removing enzyme GlnD, PII

protein GlnB, and the two component signaling system composed of histidine kinase,

NtrB, and its cognate response regulator, NtrC [97]. GlnD functions as an intracellular

sensor of nitrogen status through response to glutamine levels in the cell. Based on these

levels, GlnD can function either to add an uridyl group to GlnB or to remove it. GlnB

also serves as an intracellular nitrogen sensor for the cell based on both direct sensing of

alpha-ketoglutarate and on indirect response to glutamine levels through its modification

state [98, 99]. In low nitrogen conditions, GlnD transfers an uridyl group to GlnB

causing it to dissociate from histidine kinase, NtrB. This dissociation in turn activates

NtrB, allowing transfer of a phosphate group to NtrC. NtrC-P can then bind to upstream

activation sequences in concert with sigma-54 RNA polymerase (also called RpoN or

NtrA) to stimulate transcription of all genes containing sigma-54 promoter sites (Figure

1.5). Therefore, the proteomes of Azospirillum species grown under nitrogen fixing

25

Figure 1. 5 Levels of regulation in expression of nitrogenase structural proteins and related synthesis

proteins

Due to the energetically expensive process of nitrogen fixation, several layers of regulation are inherent in

expression of the structural components of the nitrogenase enzyme complex as well as in expression of

related proteins. GlnB and NtrB exist in a complex in the cell. When GlnD senses low nitrogen content, it

serves to uridylylate GlnB, causing a dissociation of the GlnB-NtrB complex. NtrB is then free to bind

with transcription factor NtrC-P to effect transcription of transcription factor nifA. NifA and sigma 54

binding to consensus sites upstream of the nifHDK operon then facilitates transcription of the structural

components of the nitrogenase complex as well as genes related to nitrogenase synthesis.

nifA gene

transcription

NifA

NtrC binding site

RpoN binding site

NtrC P Sigma 54

UMP

Sigma 54

RpoN binding site

NifA binding site

NtrB ADP

ATP

GlnB

NtrC

nifHDK operon

transcription

Nitrogenase Complex

GlnD NtrB

GlnB

26

conditions exhibit a different profile than those grown under non nitrogen fixing

conditions, where sigma-54 RNA polymerase is not active.

Once transcribed, the activity of the nitrogenase complex is further regulated by

post translational modifications that are also tied not only to nitrogen sensing systems

within the cell, but also to oxygen levels as oxygen is a potent inhibitor of nitrogenase

activity [98]. When combined nitrogen is not readily available, A. brasilense fixes

atmospheric nitrogen under microaerophilic conditions, preferring an oxygen content of

about 0.5% [100]. Both high levels of fixed nitrogen and higher oxygen content results

in ADP-ribosylation of the Fe protein of nitrogenase, dinitrogenase reductase, effectively

turning the enzyme off [91, 101]. A return to optimal oxygen content or unavailability

ofa sufficient concentration of fixed nitrogen results in removal of ribosylation, restoring

nitrogenase function and ultimately resulting in nitrogen fixation [98].

Much of nitrogenase experimental research has been performed in alpha-

proteobacterium Azotobacter vinelandii due to the ease with which nitrogenase can be

purified from this species [97]. Azotobacter controls transcription of the nitrogenase

protein through associations between NifA and NifL proteins [96, 97]. NifA is a sigma-

54 dependent transcriptional activator required for transcription of nitrogenase structural

genes. It has an N-terminal GAF domain, a central sigma-54-dependent activator domain

and a C-terminal DNA binding domain [97]. A variable region between the central

domain and the C-terminal DNA binding domain can either contain a CXXXXC motif,

making the NifA protein oxygen sensitive as it does in Azospirillum species, or can lack

the CXXXXC motif, making it oxygen-resistant as it is in Azotobacter species [97]. In

those species in which NifA is oxygen resistant, NifL is a partner protein to NifA,

27

expressed in a 1:1 stoichiometry [97]. NifL has an N-terminal PAS domain which

functions as a redox sensor, and a Histidine kinase-like C-terminal domain that binds

ATP and ADP, and serves to inhibit NifA activity in response to ADP and fixed nitrogen

levels [97]. In both Azotobacter and Azospirillum, NifA (and thus NifL in Azotobacter)

is constitutively expressed, but Azospirillum species do not have a nifL gene within their

genome. Expression levels of NifA in Azospirillum are moderated by oxygen and

ammonia content [91]. Activity levels of NifA are controlled through interaction with

GlnB (not NifL as in Azotobacter species), perhaps as an indicator of nitrogen status of

the cell [91]. Due to these differences, A. brasilense has served as a model for nitrogen

fixation in Azospirillum species for a number of years. In Chapter 3 of this work, we

explore the expression of nitrogenase and associated proteins for both Sp7 and Sp245

strains as well as examine the overall changes in the expression proteome during nitrogen

fixation in Azospirillum brasilense cells.

Earlier experiments suggested the presence of an alternative nitrogenase

containing an iron-only cofactor in A. brasilense strain Cd (a close relative of Sp7) [102],

but additional genetic evidence has not been forthcoming. Interestingly, the genome

sequence of strain Sp245 has indicated the presence of an alternative nitrogenase which is

proposed to use vanadium as a cofactor instead of molybdenum. A number of bacterial

species have been shown to have alternative nitrogenase systems in which the nitrogenase

uses vanadium as a cofactor for dinitrogenase (V-Fe protein), or alternatively uses only

iron [103-105]. However, most alternative nitrogenases are hierarchically expressed, and

so require total chelation of metals from the growth media, or alternatively, deletion of

the genes for nitrogenase structural components containing molybdate co-factors, in order

28

to see expression of the alternative molybdenum-free nitrogenase. With this in mind,

cultures of A. brasilense Sp245 were grown with and without added molybdate, with

vanadium added to the culture in the absence of added molybdate. Due to the difficulties

inherent with total chelation of metal from all media components and glassware, we

elected to simply add vanadium to our cultures in order to test for expression of this

alternative nitrogenase within the proteome studies. Data from both the nitrogen fixing

cultures grown with vanadium and with molybdate are included in Chapter 3 of this

work, and comparisons are made between protein expression levels in both vanadium and

molybdate nitrogen fixing cultures with those in control cultures grown under non

nitrogen fixing conditions.

Chemotaxis

Motile bacteria have a complex and highly sensitive sensory system that allows

them to sense and respond to minute changes in their environment [106]. Biased

swimming in response to chemoeffector gradients in motile bacteria is known as

chemotaxis, where increases in chemoattractants result in biased swimming up the

concentration gradient while increased concentrations of chemorepellants has the

opposite effect [106]. In a homogeneous environment, the swimming behavior of

bacteria is erratic, with regular tumbling resulting in random changes of direction [106].

The best known and most well characterized signaling system in bacteria is the E.

coli chemotaxis system [107], a large signaling complex localized to the poles of the E.

coli cell. Signal transduction in this system is mediated by a two-component system

consisting of histidine kinase, CheA, and cognate response regulator, CheY, whose

29

sequences are conserved throughout all bacterial species [106]. Sensory input to the

signaling system is accomplished through binding of chemoeffectors to membrane

spanning proteins known as methyl-accepting chemotaxis proteins, or MCPs. MCPs

contain variable periplasmic sensing domains, with conserved cytoplasmic signaling

domains [106, 108]. Clusters of MCPs at one or both poles of an E. coli cell bind

chemoeffectors and then undergo a conformational change, working cooperatively to

amplify and transmit environmental signals [108-110]. This conformational change in

MCP proteins is transmitted via docking protein CheW to histidine kinase CheA, which

then undergoes autophosphorylation [109, 110] (Figure 1.6). The phosphate group is

subsequently transferred to response regulator CheY, which then diffuses through the

cytoplasm and binds to flagellar motor switch protein, FliM, to change the direction of

flagellar rotation [109]. In E. coli, counterclockwise rotation of the flagellar bundles

allows formation of a single bundle, and results in smooth swimming, while clockwise

rotation causes the bundles to dissociate resulting in tumbling behavior and thus a

random change in direction [106].

Environments are not usually static in nature, so sensing and signaling systems

must be reset in order to respond to new and ever-changing situations. Two

modifications provide a re-set of the E. coli chemotaxis system [108, 109]. Phosphatase

CheZ dephosphorylates CheY, thus resetting the forward signaling pathway and allowing

CheY to diffuse back to the large receptor complex at the opposite pole [106]. S-

Adenosylmethionine (SAM)-dependent methyl-transferase CheR and methyl-esterase

CheB work together to modulate the level of methylation of glutamate residues within

30

Figure 1. 6 Prototypical chemotaxis pathway in E. coli Membrane spanning methyl-accepting chemotaxis proteins (MCPs) cluster at the pole of the E coli cell,

represented above by the light blue stars with tails at the pole of the cell. Binding of effectors to these

MCPs causes signal transmission to CheA histidine kinase, represented above by red ovals labeled with

“A”, through adaptor protein CheW, represented above by tan circle labeled with “W”. CheA

autophosphorylates on histidine residues and then transfers this phosphate (green circle labeled “P”) to an

aspartate residue on cognate response regulator CheY (red oval labeled “Y”), which then binds to

molecular motor protein FliM to effect a change in direction of the flagella. Adaptation to signal is

accomplished through the action of phosphate CheZ (orange circle labeled “Z”) which de-phosphorylates

CheY and causes it to diffuse back to the complex at the opposite pole from the flagella, and by

constitutively active methyltransferase CheR (light blue circle labeled “R”), and methylesterase CheB

(orange circle labeled “B”). CheR uses a methyl group from donor S-adenosylmethionine (SAM) to

methylate conserved glutamate residues in the cytoplasmic tails of the MCPs. CheB is activated to remove

methyl groups from the MCP (represented here by the methanol) by phosphorylation by CheA.

Demethylation of receptors can be determined through measuring methanol release from the cells. Levels

of methylation of the receptors determine signaling response.

SAM

B

R

A W Y

CH3OH

A

Y

Z

P P

P

CH3

CH3

CH3

FliM

31

conserved motifs in cytoplasmic tails of MCP sensory receptor proteins [106, 109]

(Figure 1.6). CheR is constitutively active, methylating glutamate residues and removing

negative charge to form methyl esters [106, 109]. Methylated receptors are more likely

to facilitate autophosphorylation of CheA upon binding of a chemoeffector, and thereby

activate the forward signaling pathway that results in a change in swimming direction

[106]. CheB functions as a methylesterase, removing methyl groups from glutamate

residues and thus restoring negative charge in the cytoplasmic tails of the MCP receptor

[109]. It has also been shown to remove amine groups from glutamine to create

glutamate, providing additional methylation sites for CheR [106]. CheB is

phosphorylated, and thus activated, by phosphotransfer from CheA. Upon CheB

phosphorylation, methylation levels of receptor cytoplasmic tails decreases, making the

signaling system less likely to react to changes in chemoeffector concentration, with the

end result being a memory function of sorts, allowing the cell to swim smoothly for a

longer period, giving time for the system to return to pre-stimulus state [109].

Many motile bacteria show a wide variety of chemotaxis components, and can

encode a number of chemotaxis operons within their genome. Multiple chemotaxis

operons are present in almost half of sequenced motile bacteria that have a CheA

homolog [111]. Although these operons are orthologous to chemotaxis operons, not all

of them are essential for chemotactic behavior, nor are they all expressed at the same

time. For instance, Rhodopseudomonas palustris encodes 3 complete chemotaxis

operons [112], but chemotactic behavior has not been studied in this species. R.

sphaeroides has 3 partial chemotaxis operons encoding 4 CheA proteins, 6 CheY, 2

CheB, and 9 MCPs. Two of the three operons are essential for chemotactic behavior,

32

with one being more highly expressed in aerobic conditions, and the other in anaerobic

conditions [106]. Pseudomonas aeruginosa, on the other hand, has 5 chemotaxis

operons, with two operons being essential for chemotaxis, while a third operon acts to

optimize chemotactic response [113]. A fourth chemotactic-like operon in P. aeruginosa

plays a role in cell aggregation and colony morphology through mediation of cyclic di-

GMP levels [114]. Hybrid molecules, such as CheV or CheC proteins of B. subtilis

containing both response regulator and docking functions [115], or phosphatase activity,

respectively, are common in more complex chemotaxis pathways [116]. Receptor

sensing molecules can be either transmembrane or wholly cytoplasmic [111].

Methylation patterns may be variable on cytoplasmic tails of MCPs depending upon the

environmental stimulus being sensed, with output response based on the particular

methylation pattern [106]. Some CheY-like response regulators control gene expression

of genes involved in developmental pathways and have no influence on microbial motion

at all [116]. Taken together, the variety of components and functions of chemotaxis

signaling pathways can enhance the signaling and response capability of the bacteria with

multiple pathways [111].

The genome sequence of A. brasilense Sp245 revealed 4 chemotaxis-like operons

and 45 MCPs, suggesting the capability of sensing multiple chemoeffectors, and

subsequently responding to signals in multiple ways. Characterization of one of these

four chemotactic pathways (herein called Che1) indicated that it has a supportive role in

chemotaxis, while also contributing to regulation of other cellular responses such as cell

length at division and clumping behavior [117, 118]. Mutants were constructed in which

either the forward signal transduction pathway (CheY) or the adaptation pathway (CheB

33

and CheR) was removed from the genome. Those mutants lacking a CheY, and thus

unable to send a signal in the forward signaling direction, exhibited a phenotype of being

considerably shorter than wild type cells and formed clumps under high aeration

conditions. In contrast, those mutants lacking both a CheB methylesterase and a CheR

methyltransferase were longer, and were impaired in clumping behavior [117]. In order

to more fully examine the effects of this chemotaxis pathway on cell function, whole cell

proteomics of these mutants were analyzed and results compared to those of wild-type

cells. The overall goal of these comparisons, presented in Chapter 4, was elucidation of

possible pathways affected by the Che1 operon.

The overall goal of this dissertation is the use of proteomic technology to

investigate the response of Azospirillum species to both changes in growth conditions,

and to manipulation of a single signal transduction pathway. The first section of this

dissertation involves optimization of mass spectrometer parameters in order to maximize

protein identifications while minimizing the duty cycle of the mass spectrometer. In the

second section of this dissertation, the optimized mass spectrometry parameters are used

in the investigation of the total proteome expression of two different strains of newly

sequenced bacteria, Azospirillum brasilense, grown under both nitrogen fixing and non-

nitrogen fixing conditions. Further investigations include examination of the effect of

mutations introduced in components of a chemotaxis-like operon on the proteome

expression of A. brasilense species Sp7. Mass spectrometry based proteome analyses

comparing bacterial isogenic mutant strains lacking components involved in signal

transduction and adaptation pathways were compared, and results are presented in

Chapter 4. And finally, a set of vectors with broad host range expression capability that

34

offers the advantage of ligation-free cloning and choice of C-terminal tandem affinity

tags was developed, and is presented in Chapter 5. This set of vectors can be used to

characterize the function of proteins found to be of interest in nitrogen fixation or to be a

consequence of alteration in the Che1 chemotactic pathway.

35

Chapter 2. Optimization of Tandem Mass Spectrometry

Methods

Introduction

As discussed in Chapter 1, shotgun proteomics has become a very powerful and

commonly used methodology for investigating proteome structure, whether a survey of

all proteins present within a cell lysate, an investigation of proteins and their interacting

partners, or quantitative or semi-quantitative comparisons of concentrations of proteins

present under different conditions [4, 40, 119, 120]. Use of a “bottom up” tandem mass

spectrometry (MS/MS) approach in which proteins are first digested into peptides and

then peptides are detected within a mass spectrometer allows for elucidation of the

sequences of peptides, and subsequent identification of proteins present in the entire

protein complement of a cell [72]. In tandem mass spectrometry (MS/MS) experiments

performed in an ion trapping mass spectrometer, the mass spectrometer isolates a selected

ion through application of appropriate voltages, with a specific mass-to-charge (m/z)

window being selected for isolation. The isolated parent then undergoes fragmentation

through collision with a target gas to create a set of fragment ions that are scanned out of

the trap to yield an MS/MS spectrum, as shown in figure 1.2. The spacing of peaks based

on the difference in mass of individual side chains of amino acids within the fragment ion

spectrum provides information on the amino acid sequence of the parent ion. Single or

even multidimensional chromatography in line with the mass spectrometer provides

separation of extremely complex mixtures of peptides such that subsets of peptides are

eluted and injected directly into the mass analyzer. This on-line separation allows for

detection of less abundant peptides in a complex mixture, resulting in acquisition of a

36

greater number of peptides per individual protein in the mixture, and thereby increasing

the depth of analysis. Database searching algorithms such as SEQUEST [3] or Mascot

[53] can then match this experimental spectrum to theoretical spectra derived from the

database sequences, providing a score for the top matches. Other programs such as

DTASelect [54] filter and sort this output data to yield a final output of peptide

identifications which in turn contribute to individual protein identifications.

Identification of the correct peptide sequence from this raw spectrum depends

upon the ability of a database search algorithm such as SEQUEST [3] to compare a sub-

optimal experimental spectrum to a theoretical computer-generated (and thus “optimal”)

spectrum and find the very best match available. This raises the question of what factors

contribute to making an “optimal” experimental spectrum. What spectral qualities will

contribute to the maximum number of true positive identifications while at the same time

ensuring the minimal number of false identifications? Certainly, the score thresholds

specified in filtering programs like DTASelect [54] for accepting a positive identification

can be changed such that the number of false positives is below an acceptable threshold

[62]. However, raising score thresholds comes at the cost of rejecting a small

proportion of peptides that are true positives along with those that are falsely deemed to

be positive identifications. Additionally, different scoring algorithms, such as MASPIC

[121] or other search algorithms such as Mascot [53], DBDigger [7], MyriMatch [6], or

InsPect [122] can be used and the results compared to improve the confidence level of

positive identifications. Interestingly, earlier studies have indicated that Mascot and

SEQUEST perform comparably when used on spectral data acquired on a linear ion

trapping quadrupole instrument [119].

37

The above mentioned options are certainly viable, but the parent ion and

fragmentation spectra collected by each mass spectrometer provide the raw data input to

these search algorithms. If an instrument is not behaving optimally, manipulation of the

search algorithm can only yield minimal gains due to a dearth of good quality spectra.

Parameters by which the spectral quality may be evaluated include the ion intensity of the

parent ion and the intensity and distribution of the fragment ions resulting from collision-

induced dissociation of this parent ion. A number of experimental parameters can

contribute to the final quality of the spectra obtained. Sample preparation methods

resulting in low concentrations of proteins can result in less than optimal spectra due to

the low intensity or absence of fragment ion peaks and can interfere with relative

quantitation of sample amounts between two different growth states [15]. Separation

methods can either reduce or increase the concentration of peptides available to the mass

analyzer, but can result in the presence of co-eluting peptides that may be sampled by the

mass analyzer in a given sampling cycle [123]. Instrument parameters in an ion trap

mass spectrometer such as the ion injection time, collision energy, scan rate, number of

averaged scans (microscans) that make up the raw spectrum, and width of the mass-to-

charge (m/z) isolation window can all affect the quality of the spectra obtained. Longer

ion injection times can result in more ions entering the trap, which in turn can result in

greater ion intensities for the parent ion, perhaps leading to better fragmentation spectra

[57, 58]. However, a longer ion injection time can also result in too many ions in the trap

causing a decrease in trapping efficiency due to space charge effects [124]. Low scan

rates result in greater ion intensities for high m/z ions and low intensities for low m/z

38

ions, while the opposite is true for high scan rates [57]. Collision energy has a variable

effect on MS/MS spectra that is dependent on the settings of other parameters [57].

Data acquisition in a tandem mass spectrometry experiment in an ion trap

instrument begins with filling the trap with ions being sprayed from the ion source at that

point in the chromatographic run. The instrument software ramps voltages so that ions of

decreasing m/z values become unstable and are scanned out of the trap to a detector. It

then catalogs those ions that are present in the full MS scan. When an instrument is

operated in data-dependent mode, a parent ion is chosen for isolation based on user-

selected criteria addressing the issues of whether the ion has been scanned recently within

a specified amount of time (termed dynamic exclusion), and how much of a window on

either side of the peak of the ion is to be isolated. Each parent ion is made up of a series

of peaks resulting from isotopologues including one or more “heavy” isotopes of each

element. For instance a population of ions will have some percentage that contains one

13C atom with all the rest of the carbon atoms being

12C, thus this small population of

ions will be mass shifted by 1 Da. Likewise, some percentage will contain two 13

C

atoms, shifting the mass by 2 Da, and so on. Larger molecules thus have more possible

isotopologues, resulting in a wide distribution of individual isotopologue peaks with an

approximate Gaussian distribution, where the isotopologue with the most abundant mass

is at the centroid of the peak. The goal of isolating a parent ion for fragmentation

includes isolating the entire isotopic packet of that parent ion, accomplished through

setting an isolation m/z “window” centered around the central mass peak. Data is

typically collected using preset isolation window widths of 2 or 3 m/z units (1 or 1.5 m/z

respectively on each side of the peak), which can be set within the software controlling

39

the mass spectrometer. A wider isolation width of 5 m/z units has the potential to yield

more total ions, more intensity and anecdotally better MS/MS spectra (Verberkmoes,

personal communication). However, co-isolation of other peptides may be a potential

negative consequence of the larger mass-to-charge isolation window (McDonald,

personal communication). In our lab an increase in identifications has been noted with a

larger mass-to-charge isolation window of 5 m/z. Here we test isolation widths of 2, 3, 4

or 5 m/z to ascertain the effect of varying this parameter on the number of peptide and

protein identifications.

Once the parent ion is isolated in the trap, fragmentation via collision induced

dissociation (CID) yields an experimental fragmentation spectrum that can be matched to

predicted spectra derived from a database. A number of studies have been done to

investigate the reproducibility of spectra between replicate runs [59, 125]. Mass spectra

collected from the same peptide in different replicate runs may have widely variant

intensities for each peak collected, and may vary widely in the presence of low

abundance fragment spectral peaks [59]. The differences in spectral quality due to

variable ion intensities can be compensated for by collecting a number of spectra of the

same ion (microscan), and averaging all of these spectra together to produce a final single

raw output spectrum. When acquiring data, the mass spectrometer takes one or more

individual “microscans” of repeated fills of the trap. The spectra obtained from each

microscan are then averaged to create the final displayed MS scan as illustrated in Figure

2.1. More microscans result in better representation of the ions present in the trap, which

should ultimately result in a better probability of obtaining higher quality spectra, with

40

Figure 2. 1 Representation of spectral averaging (number of microscans) to obtain a final observed spectrum The ion trap is filled and a parent ion is isolated and subjected to collision induced dissociation to create a fragment ion spectrum which is recorded by

instrument software, but not displayed. This process is repeated a number of user-specified times and the peak intensities and positions of the individual

spectra are averaged to create a final fragment ion spectrum that is displayed and recorded. Increasing the number of microscans can result in a cleaner

spectrum with greater signal to noise ratio and a greater number of fragment ions, but the trade-off is the amount of time required to do multiple scans.

o p t i m i za t i o n _ m i c r o s c a n - 2 b # 3 2 6 4 R T : 7 5 . 5 9 A V : 1 N L : 2 .8 7 E 7

T : + c N S I d F u l l m s 2 5 6 6 . 3 6 @ 3 5 . 0 0 [ 1 4 5 .0 0 - 1 7 1 0 . 0 0 ]

4 0 0 4 5 0 5 0 0 5 5 0 6 0 0 6 5 0 7 0 0 7 5 0 8 0 0 8 5 0 9 0 0 9 5 0 1 0 0 0 1 0 5 0 1 1 0 0 1 1 5 0 1 2 0 0 1 2 5 0 1 3 0 0 1 3 5 0 1 4 0 0

m /z

0

5

1 0

1 5

2 0

2 5

3 0

3 5

4 0

4 5

5 0

5 5

6 0

6 5

7 0

7 5

8 0

Re

lativ

e A

bund

an

ce

7 6 7 .2 0 9 0

8 1 6 . 1 3 7 1

5 8 1 .2 9 2 0

9 2 9 . 2 4 1 9

4 6 5 .3 0 9 1

4 5 1 . 3 5 9 8

7 3 1 .9 5 9 4

5 5 8 .3 6 4 0

4 9 4 .2 7 1 67 0 1 .2 4 2 1

5 4 4 . 2 9 9 6 6 0 1 . 6 7 3 66 6 6 .4 3 0 5

9 1 2 . 1 7 2 96 4 3 . 3 5 2 1

8 8 0 . 3 4 4 6 1 1 6 4 . 3 8 7 21 0 1 5 . 9 7 3 0 1 1 1 5 .2 8 8 1 1 2 5 0 .0 0 4 6 1 3 5 0 . 1 6 0 29 7 2 . 8 5 5 3

O p ti m i z a t i o n _ M i c ro s c a n - 5 # 1 8 6 8 R T : 7 6 .0 5 A V : 1 N L : 1 .9 8 E 7T : + c N S I d F u l l m s 2 5 6 6 .2 5 @ 3 5 . 0 0 [ 1 4 5 .0 0 - 1 7 1 0 .0 0 ]

4 0 0 4 5 0 5 0 0 5 5 0 6 0 0 6 5 0 7 0 0 7 5 0 8 0 0 8 5 0 9 0 0 9 5 0 1 0 0 0 1 0 5 0 1 1 0 0 1 1 5 0 1 2 0 0 1 2 5 0 1 3 0 0 1 3 5 0

m /z

0

2

4

6

8

1 0

1 2

1 4

1 6

1 8

2 0

2 2

2 4

2 6

2 8

3 0

3 2

3 4

3 6

3 8

4 0

4 2

4 4

4 6

4 8

5 0

5 2

5 4

Re

lativ

e A

bu

ndan

ce

7 6 7 . 2 3 9 7

5 8 1 .2 5 1 5

8 1 6 .1 8 4 49 2 9 .2 8 8 3

5 6 0 .1 3 2 84 6 5 .7 9 4 9 7 3 2 .1 1 3 4

4 5 1 .1 2 8 0

6 0 1 .8 3 6 97 0 1 .1 8 3 74 9 4 .1 8 8 6

5 4 4 . 0 7 7 0

9 9 6 .1 5 7 26 6 6 .4 2 8 2

1 3 4 9 .1 7 3 39 1 2 .0 5 6 8 1 1 1 6 .6 4 7 96 1 6 .2 8 2 2 8 6 4 .1 6 0 8 1 2 3 3 . 1 1 8 29 6 8 . 4 9 9 0 1 0 6 9 .3 6 2 8 1 1 7 1 .2 3 6 8 1 3 6 4 . 2 3 4 11 2 7 7 .3 1 9 3

o p t i m i z a t i o n _ m i c r o s c a n - 3 # 2 5 5 8 R T : 7 5 . 9 6 A V : 1 N L : 3 . 4 6 E 7

T : + c N S I d F u l l m s 2 5 6 6 . 3 3 @ 3 5 . 0 0 [ 1 4 5 . 0 0 - 1 7 1 0 . 0 0 ]

4 0 0 4 5 0 5 0 0 5 5 0 6 0 0 6 5 0 7 0 0 7 5 0 8 0 0 8 5 0 9 0 0 9 5 0 1 0 0 0 1 0 5 0 1 1 0 0 1 1 5 0 1 2 0 0 1 2 5 0 1 3 0 0 1 3 5 0 1 4 0 0

m / z

0

2

4

6

8

1 0

1 2

1 4

1 6

1 8

2 0

2 2

2 4

2 6

2 8

3 0

3 2

3 4

3 6

3 8

4 0

4 2

4 4

4 6

4 8

5 0

5 2

5 4

5 6

5 8

6 0

Rela

tive A

bu

nda

nce

7 6 7 . 2 7 7 8

8 1 6 . 1 9 4 65 8 1 . 2 5 2 7

9 2 9 . 2 8 7 4

4 6 5 . 1 3 3 5

7 3 2 . 4 0 1 2

5 5 8 . 2 2 0 06 0 2 . 2 8 4 9

4 9 4 . 2 3 5 8

5 3 5 . 4 9 8 8 7 0 1 . 3 0 6 36 6 6 . 4 3 2 19 1 1 . 2 8 0 9

9 9 7 . 7 1 9 06 2 1 . 1 8 7 5 1 0 5 6 . 0 9 1 18 8 6 . 1 0 7 11 3 4 7 . 2 8 6 61 1 6 3 . 6 4 0 9 1 2 7 6 . 8 4 0 31 2 3 3 . 7 4 4 1

+

Final observed spectrum = Average of microscan spectra

41

better ion intensities and a greater number of fragment ion peaks. Averaging a number of

microscans normalizes the intensities of each ion present in the trap at a given time for

both parent ion (MS) scans and for resulting fragment ion (MS/MS) scans [60], making

the experimental spectrum more closely resemble the theoretical spectrum used by

database search algorithms. A more populated fragmentation spectrum resulting from an

increased number of microscans can potentially lead to a more confident identification as

the experimental spectrum approaches a more optimal appearance.

However, earlier studies using different numbers of microscans have led to

conflicting results. When separating peptides from a single protein bovine serum albumin

(BSA) digest and using an ion trapping quadrupole mass spectrometer (LCQ, Thermo-

Finnigan) for mass analysis, Wenner et al [60] determined that a window of optimal

microscan numbers existed, with either 3, 5, 7 or 9 microscans resulting in a higher

number of peptide identifications than those obtained using either higher or lower

numbers of microscans. Experiments done by Venable et al [59] also indicated that

increasing the number of microscans when acquiring tandem mass spectrometry data

from protein mixtures on an LCQ leads to less total identifications of peptides, but yields

a better separation of true positive identifications from false positive identifications.

However, subsequent studies by Moberg and colleagues indicate that increasing the

number of microscans has no significant effect on the identifications made when

collecting tandem mass spectrometry data on a linear ion trapping quadrupole (LTQ)

instrument [57].

42

Instrument software allows a choice of the number of microscans performed to

make up the final spectrum. Previous data from the group indicates that performing five

microscans for each tandem mass spectrum on the LCQ results in better data quality than

performing three microscans or one microscan (data not shown). Data quality output

from increasing microscan numbers in the LTQ have not been investigated in our group.

As stated earlier, an increased number of microscans should yield better ion statistics and

thus should improve both sensitivity and specificity of data-dependent LC/MS-MS

analysis of peptides. This increased sensitivity and specificity should in turn decrease the

number of false negatives and false positives returned by our data analysis pipeline.

However, performing an increased number of microscans requires an increased amount

of time, and thus results in collection of fewer total spectra with the attendant possibility

of missing some peptides. Based on previous data from the group, we compare data

quality from two, three or five tandem mass spectrometry microscans on the LCQ ion

trapping quadrupole instrument. The LTQ linear two dimensional quadrupole ion trap

mass spectrometer has larger ion storage capacity and higher trapping efficiency, trapping

20-40 times as many ions as does the LCQ [48]. A higher number of ions in the trap

should provide better ion statistics, and thus better data quality [48]. Here, for the LTQ,

we compare the data quality obtained from one microscan versus two or three

microscans.

The overall goal of experiments in this chapter is to optimize the mass

spectrometry operation parameters of isolation width m/z window and number of

microscans employed to obtain an MS/MS spectrum in order to maximize the number of

true positive identifications. A systematic examination of results obtained when tandem

43

mass spectrometer parameters are varied will lead to a more optimized operation of the

mass spectrometer, which is essential for maximizing the number of identifications made

in any given experiment. Subsequent use of these parameters in proteomics experiments

performed in chapters 3 and 4 of this work will allow for more accurate determination of

the protein components present in a given proteomic experiment.

Materials and Methods

Chemicals

All chemicals were obtained from Sigma Chemical Company (St. Louis, Mo)

unless otherwise stated. Acetonitrile, and HPLC-grade water were purchased from

Burdick and Jackson (Muskegon, MI ), 98% formic acid (FA) from EM Science (an

affiliate of MERCK KgaA, Darmstadt, Germany), and sequencing-grade trypsin from

Promega (Madison, WI).

Preparation of extended protein standard mix (EPSM)

The extended protein standard mixture was prepared as described earlier [126].

Briefly, twenty proteins listed in Table 2.1 were mixed in equimolar amounts and

suspended in 6 M guanidine hydrochloride with 10 mM DTT. This stock solution was

frozen at -80°C for later digestion with trypsin as related below. Once digested, the

EPSM digest was concentrated by vacuum centrifugation to a final concentration of 1

mg/ml, aliquoted into individual tubes, and stored at -80°C for later analysis.

44

Table 2. 1 Proteins included in the Extended Protein Standard Mix (EPSM)

Protein Organism

Carbonic Anhydrase II Bos taurus

Conalbumin (ovotransferrin) Gallus gallus

Concanavalin A Canavalia ensiformis

Cytochrome c Bos taurus

Deoxyribonuclease I Bos taurus

Lysozyme c Gallus gallus

Beta-lactoglobulin B Bos taurus

Ribonuclease A Bos taurus

Thyroglobulin Bos taurus

Serum albumin Homo sapiens

Serum albumin Bos taurus

Alcohol dehydrogenase E Equus caballus liver

Alcohol dehydrogenase I Saccharomyces cerevisiae

Alpha-amylase Bacillus subtilis

Beta-amylase Ipomoea batatas

Apomyoglobin Equus caballus

Hemoglobin A Equus caballus

Hemoglobin B Equus caballus

(Apo)-transferrin Bos taurus

45

Preparation of R. palustris standard proteome

Cell Growth

R. palustris was grown photoheterotrophically as described earlier [30] under

anaerobic conditions in 1.5 L of defined mineral medium with 12.5 mM sodium

phosphate, 12.5 mM potassium phosphate,10 mM succinate, 0.1% ammonium sulfate and

2 mg/L p-aminobenzoic acid. Nitrogen gas replaced the air in the head space, and

cultures were grown at 30°C with a light source to mid-log phase (OD660nm ~ 5.0) while

mixing the culture with a stir bar. Cells were harvested by centrifugation at 5000 xg in an

S-17 rotor in a Sorvall centrifuge for 20 minutes at 4°C, washed twice with cold 50 mM

Tris buffer, and pellets were frozen at -80°C for later lysis.

Cell lysis

Cell pellets were thawed and resuspended in ice-cold buffer (50 mM Tris, 10mM

CaCl2, pH 7.4). After resuspension, cells were lysed by sonication (Misonix Sonicator

4000, Newtown, CT, 4 x 30 sec, 40% power, with 30 second cooling periods) and cell

debris cleared by centrifugation at 18,000 xg for 10 minutes. Total protein concentration

was then determined through bicinchoninic acid (BCA) assay (Pierce Protein Research, a

division of Thermo Scientific) following manufacturer’s protocols. Supernatant was then

aliquoted and frozen at -80°C.

Trypsin Digestion

Trypsin digestion was done following protocols described earlier for S. oneidensis

with minor modifications [126]. Briefly, one proteome aliquot (approximately 6 mg of

protein) was denatured by addition of 6M Guanidine to the lysate with subsequent

46

incubation at 60°C for 1 hour. Guanidine was then diluted 6-fold through addition of 5

volumes of 50 mM Tris/10 mM CaCl2 pH 7.4, and 20 µg sequencing-grade trypsin was

then added for every mg of protein. Digestion was done for 18 hours at 37°C with

rotation. More sequencing-grade trypsin was added (20 µg/mg protein) and digestion

continued at 37°C with rotation for an additional 5 hours. A final reduction step was

done through addition of 20 mM DTT followed by incubation for 1 hour at 60°C.

Samples were then desalted using a Sep-Pak plus C-18 solid-phase extraction column

(Waters, Milford, MA), and extracted with acetonitrile/0.1% FA. Samples were

concentrated through centrifugal evaporation (Savant Instruments, Holbrook, NY), and

solvent-exchanged to water with 0.1% FA. Samples were then filtered, aliquoted and

frozen at -80°C until MS analysis.

LC/LC-ES-MS/MS Analysis:

EPSM Chromatography

Six technical replicates of EPSM were analyzed. Stock solution of the digested

EPSM at a concentration of 1 mg/ml was diluted 1:10 in Switchos Buffer A (100%

HPLC-grade water, 0.1 % formic acid (FA)) for a final concentration of 0.1 mg/ml. Tips

were pulled using a P-2000 laser puller (Sutter Instruments, Novato, CA) from 100µ ID

fused silica to a 5µ tip opening, and packed via pressure cell (New Objective, Woburn,

MA) with 15cm Aqua C-18 5µm reverse phase (RP) packing material (Phenomenex,

Torrance, CA) to form the resolving column. Fifty microliters of the 0.1 mg/ml solution

was injected via FAMOS autosampler onto a reverse-phase trapping column. Peptides

were then eluted from the trapping column to the resolving column, which was placed

47

directly in-line with either an LCQ-DecaXP quadrupole ion trap mass spectrometer

(Thermo- Fisher Scientific, Waltham, MA) or an LTQ linear ion trap mass spectrometer

(Thermo-Fisher Scientific, Waltham, MA). Peptides were eluted from the resolving

column by gradient elution from 100% Buffer A (100% HPLC-grade water, 0.1 % FA) to

100% Buffer B (70% acetonitrile, 0.1% FA) over a time period of 120 minutes.

Experiments were performed in random order, with blank chromatography runs between

each experimental condition in order to minimize carry-over between runs.

Proteome Chromatography

Proteome samples were analyzed in triplicate via 5-step multidimensional protein

identification technique (MudPIT) [17, 19] performed with an Ultimate HPLC (LC

Packings, a division of Dionex, San Fransisco, CA) coupled to either an LCQ-Deca

quadrupole ion trap mass spectrometer (Thermo- Fisher Scientific, Waltham, MA) or an

LTQ linear ion trap mass spectrometer (Thermo-Fisher Scientific, Waltham, MA). A

triphasic split column was used for separation of peptides. The front resolving column

was constructed as described for the EPSM above and was placed in-line with the mass

spectrometer. Back columns consisted of 150µ ID fused silica, and were loaded first

with 3 cm of Luna 5µm strong cation exchange (SCX) resin (Phenomenex, Torrance,

CA) followed by 3cm of Aqua 5 µm C-18 reverse phase resin (Phenomenex). Proteome

aliquots, each 10µL of approximately 1 mg/ml concentration, were loaded directly onto

the back column via pressure cell. The back column was then coupled to the front

resolving column.

48

Five individual chromatography steps were set up using Xcaliber software.

Initial desalting and elution of peptides to the strong cation exchange resin was done with

a 120 minute gradient elution of 100% buffer A (95% water, 5% acetonitrile, 0.1% FA)

to 100% buffer B (70% acetonitrile/ 0.1% FA). Three subsequent MudPIT [17, 20]

chromatography steps included 2 minute salt pulses of 10, 20, and 40% buffer C (500

mM ammonium acetate, 0.1% FA) to elute subsets of peptides from SCX resin to the

resolving column. These subsets of eluted peptides were then resolved by 120 minute

gradient elution from 100% buffer A to 100% Buffer B. The final MudPIT

chromatography step consisted of a 20 minute salt pulse of 100% buffer C followed by a

120 minute gradient from 100% buffer A to 100% buffer B. Columns were washed with

10 minutes of 100% buffer B followed by 50% buffer B, then re-equilibrated to 100%

buffer A before each run.

Mass spectrometry

Data acquisition was under the control of Xcaliber software (Thermo Fisher

Scientific), with data being collected in data-dependent mode. Dynamic exclusion was

enabled, with repeat count of 1, repeat duration of 0.5 minutes and exclusion duration of

1.0 minute. Mass spectrometry parameters that were held constant in all experiments are

listed in Table 2.2. For the LCQ, a full MS scan of ions with mass-to-charge ratios of

400 to 2000 m/z was performed, followed by 3 data-dependent tandem mass

spectrometry (MS/MS) scans, while one full MS scan followed by 6 data-dependent

MS/MS scans were performed for the LTQ. Tandem mass spectrometry (MS/MS)

parameters were varied according to the 6 experimental conditions listed in Table 2.3.

49

Table 2. 2 Mass spectrometry operation parameters held constant

LCQ parameters LTQ parameters

Normalized Collision energy 35.0% 35.0%

No. scan events 4 – 1 full, 3 data dependent 7 – 1 full, 6 data dependent

Activation Q 0.250 0.250

Activation time 30.00 30.00

No. parental microscans 2 1

Dynamic Exclusion Enabled Enabled

Repeat count 1 2

Repeat duration 0.50 min 0.30 min

Exclusion list size 50 100

Exclusion duration 60 sec 180 sec

Table 2. 3 Variable tandem mass spectrometer experimental parameters

LCQ variable parameters LTQ variable parameters

Experiment Isolation

width

No. of Microscans Experiment Isolation

width

No. of Microscans

1 2 m/z 3 1 2 m/z 2

2* 3 m/z 3 2

* 3 m/z 2

3 4 m/z 3 3 4 m/z 2

4 5 m/z 3 4 5 m/z 2

5 3 m/z 2 5 3 m/z 1

6 3 m/z 5 6 3 m/z 3

50

Based on results from experiments done on extended protein mixture, only experimental

conditions 2, 4, and 5 from Table 2.3 were performed on proteome mixtures. MS/MS

parameters were varied randomly for each run in order to ensure statistical validity to

data collected and to minimize effects of instrument variability.

Data Analysis

Data from 6 independent EPSM replicates and 3 proteome replicates for each

experimental condition was pooled and searches were performed on both pooled data and

individual independent experimental runs using both SEQUEST [3] and DBDigger [7].

EPSM data was searched against two different databases, one containing the protein

sequences for the 20 proteins included in the extended protein standard mix (EPSM) plus

common contaminants with the R. palustris proteome database as a distractor. The

second database contained reversed sequences of the EPSM plus common contaminants

appended to the forward sequences. Proteome data was searched against a concatenated

forward and reverse R. palustris database including 36 contaminants

(http://compbio.ornl.gov/rpal_proteome/databases). Data was then filtered and sorted

using DTASelect [54] with the following parameters: Xcorr thresholds of 1.8 for ions

with a +1 charge state, 2.5 for +2 ions, 3.5 for +3 ions, deltaCN of 0.08, requiring 2

peptides for positive identification of a protein. Using results from the reverse database

searches, threshold levels were then determined which would yield a 95% confidence

level for each experimental condition, and data was re-filtered.

Statistics were computed on both the percentage of identifications (number of true

identifications/ total spectra collected) and the number of true identifications using

51

individual data files. Means for each individual experimental category (6 replicates in

each category) for both percentage of total peptide identifications and number of true

identifications were calculated, and compared using one-way ANOVA tests. Pairwise

comparisons of mean numbers of true identifications for the EPSM data were performed

with Tukey’s F tests to determine which means were significantly different from the

others. All statistical analysis was performed using SAS data analysis software, with p-

values less than 0.05 being considered significant.

ROC (Receiver Operating Characteristics) curves were generated from the EPSM

data using R statistical package. Data for each set of experiments was pooled and files of

true positive identifications and true negative identifications were created from the

DTASelect.txt output files for each experiment using in-house software. Receiver

Operating Characteristc (ROC) curves were created by plotting the number of true

negative identifications versus the number of true positive identifications for peptides

identified. Individual plots were done for those peptides identified as having charge

states of plus one, plus two and plus three. Plots were not normalized, but were instead

used for reporting the maximum number of known true identifications for each data set.

Isolation window widths of two, three, four, and five m/z were compared in each ROC

curve. Microscan ROC curves compared one, two, and three microscans for the LTQ

data, and two, three and five microscans for the LCQ.

Results

Initial qualitative analysis, done by comparing the highest scoring spectra among

the isolation width experiments and among the microscan experiments indicated no

52

discernible visual difference in spectral quality among MS/MS spectra acquired using

different isolation width windows or those acquired with different microscan numbers for

either mass spectrometer tested. Representative spectra from the peptide sequence

DLILQGDATTGTDGNLELTR derived from protein conconavalinA collected using

different isolation widths and different numbers of MS/MS microscans are shown in

Figures 2.2 and 2.3, respectively. Examination of these spectra shows similar intensity

levels for most abundant peaks, similar signal-to-noise ratios and a similar population of

fragment ion peaks, indicating that changing isolation width windows or number of

microscans does not have a significant effect on the appearance of the individual

spectrum.

Effect of varying data acquisition isolation width

LCQ quadrupole 3D ion trapping mass spectrometer results

Receiver Operating Characteristic (ROC) curves, discussed more thoroughly

below, were created by establishing a table containing the number true positive

identifications and the number of false positive identifications for a given threshold level

based on those proteins known to be in the EPSM mixture. Curves were created by

plotting the number of false positive identifications on the x-axis versus the number of

true identifications on the y-axis. Each point on the curve corresponds to a plot of these

numbers at a given threshold level. Examination of ROC curves for data collected on the

LCQ mass spectrometer suggests an isolation window width of 2 m/z units is best for

those peptides having a +2 or +3 charge state, while for peptides having a +1 charge state

53

Figure 2. 2 Representative fragmentation spectra from peptide sequence

DLILQGDATTGTNLELTR derived from protein conconavalinA collected using different isolation

widths NL = normalization level, Iso-width = isolation width. Examination of the fragmentation spectra collected

at each of the above isolation widths on the LCQ shows a well-populated spectrum with similar intensity

levels and signal-to-noise ratios for each of the experimental spectrum.

L C Q _ I s o w i d t h _ 4 _ 6 _ 1 4 _ 0 6 # 2 0 1 6 R T : 8 6 . 9 3 A V : 1 N L : 2 . 5 3 E 5

T : + c N S I d F u l l m s 2 1 0 5 2 . 0 6 @ 3 5 . 0 0 [ 2 7 5 . 0 0 - 2 0 0 0 . 0 0 ]

5 0 0 6 0 0 7 0 0 8 0 0 9 0 0 1 0 0 0 1 1 0 0 1 2 0 0 1 3 0 0 1 4 0 0 1 5 0 0 1 6 0 0 1 7 0 0 1 8 0 0 1 9 0 0 2 0 0 0

m / z

0

2

4

6

8

1 0

1 2

1 4

1 6

1 8

2 0

2 2

2 4

2 6

2 8

3 0

3 2

3 4

3 6

3 8

4 0

4 2

4 4

4 6

4 8

5 0

5 2

Re

lative

Ab

und

an

ce

1 0 7 5 . 4 2 3 88 8 1 . 5 7 9 7 1 1 7 6 . 4 6 4 4

1 5 2 0 . 6 2 7 2

9 2 7 . 3 5 5 8 1 2 7 7 . 3 3 1 1

8 2 6 . 2 1 8 1

8 0 2 . 3 5 3 0

1 0 1 0 . 2 0 0 2

5 8 2 . 9 0 7 71 3 4 8 . 7 0 6 8

1 5 8 6 . 5 6 9 15 1 8 . 1 4 3 11 4 6 4 . 4 1 2 8

1 9 2 8 . 5 7 5 7

1 8 1 0 . 5 6 8 61 7 1 5 . 3 6 8 26 3 1 . 4 4 3 1

1 2 4 2 . 7 6 4 99 9 2 . 0 2 2 9

6 7 4 . 6 6 8 7 7 4 5 . 6 0 1 9

4 5 5 . 0 5 0 7 1 4 2 5 . 2 8 5 2

L C Q _ I s o w i d t h _ 2 B _ 6 _ 1 4 _ 0 6 # 1 9 1 0 R T : 8 3 . 2 6 A V : 1 N L : 1 . 1 2 E 5

T : + c N S I d F u l l m s 2 1 0 5 2 . 2 6 @ 3 5 . 0 0 [ 2 7 5 . 0 0 - 2 0 0 0 . 0 0 ]

5 0 0 6 0 0 7 0 0 8 0 0 9 0 0 1 0 0 0 1 1 0 0 1 2 0 0 1 3 0 0 1 4 0 0 1 5 0 0 1 6 0 0 1 7 0 0 1 8 0 0 1 9 0 0 2 0 0 0

m / z

0

5

1 0

1 5

2 0

2 5

3 0

3 5

4 0

4 5

5 0

5 5

6 0

6 5

7 0

7 5

8 0

8 5

9 0

9 5

1 0 0

Re

lativ

e A

bu

nd

an

ce

1 0 7 5 . 4 3 1 9

1 2 7 7 . 4 5 2 9

1 5 2 0 . 5 8 6 7

1 1 7 6 . 2 7 6 9

1 3 4 8 . 7 0 7 0

1 5 8 5 . 6 0 4 5

1 0 4 3 . 0 3 6 1

9 1 7 . 3 8 9 2

9 9 2 . 6 3 0 4 1 6 4 8 . 6 8 4 3 1 9 2 9 . 7 4 4 65 1 8 . 3 0 5 3 1 1 9 8 . 7 2 2 4

8 0 2 . 3 4 9 9

1 7 1 5 . 7 7 9 51 4 7 2 . 5 6 5 4

1 8 2 8 . 7 2 3 41 1 1 8 . 1 0 4 58 8 1 . 2 9 3 8

1 4 3 5 . 3 7 6 77 7 0 . 1 6 7 5 1 7 6 1 . 9 5 7 55 8 3 . 2 5 1 5

4 3 8 . 7 7 7 06 7 8 . 0 3 0 6

1 8 6 7 . 9 9 0 5

L C Q _ I s o w i d t h _ 3 B _ 6 _ 1 4 _ 0 6 # 1 2 8 7 R T : 5 5 . 4 5 A V : 1 N L : 5 . 6 6 E 6T : + c N S I d F u l l m s 2 1 0 5 2 . 3 7 @ 3 5 . 0 0 [ 2 7 5 . 0 0 - 2 0 0 0 . 0 0 ]

5 0 0 6 0 0 7 0 0 8 0 0 9 0 0 1 0 0 0 1 1 0 0 1 2 0 0 1 3 0 0 1 4 0 0 1 5 0 0 1 6 0 0 1 7 0 0 1 8 0 0 1 9 0 0 2 0 0 0

m / z

0

5

1 0

1 5

2 0

2 5

3 0

3 5

4 0

4 5

5 0

5 5

6 0

6 5

7 0

7 5

8 0

8 5

9 0

9 5

1 0 0

Re

lative A

bun

da

nce

1 0 7 5 . 4 4 7 31 1 7 6 . 4 9 1 2

1 2 7 7 . 6 5 6 3

1 0 4 3 . 3 3 8 4

1 5 2 0 . 7 6 1 7

8 0 2 . 3 9 0 7

8 2 6 . 4 5 7 8

1 0 1 0 . 3 6 7 9

8 8 2 . 0 0 0 61 3 4 8 . 6 2 6 29 2 7 . 3 5 4 4 1 5 8 5 . 5 5 2 2

1 4 6 3 . 4 3 7 0

1 8 2 7 . 7 6 1 09 9 2 . 2 1 8 4

1 7 1 4 . 7 4 6 6 1 9 2 9 . 1 7 9 9

5 8 3 . 0 3 2 0 6 9 6 . 1 7 0 55 1 8 . 3 2 2 3

4 7 2 . 3 3 8 11 8 9 2 . 2 3 1 4

1 7 6 1 . 7 7 8 6

Iso-width 2 m/z

Iso-width 3 m/z

Iso-width 4 m/z

NL = 5.6E6

NL= 1.12E5

NL=5.3E5

L C Q _ i s o w i d t h 5 - 6 - 1 4 - 0 5 # 2 1 3 5 R T : 9 0 . 9 0 A V : 1 N L : 1 . 6 3 E 5

T : + c N S I d F u l l m s 2 1 0 5 2 . 3 3 @ 3 5 . 0 0 [ 2 7 5 . 0 0 - 2 0 0 0 . 0 0 ]

5 0 0 6 0 0 7 0 0 8 0 0 9 0 0 1 0 0 0 1 1 0 0 1 2 0 0 1 3 0 0 1 4 0 0 1 5 0 0 1 6 0 0 1 7 0 0 1 8 0 0 1 9 0 0 2 0 0 0

m / z

0

5

1 0

1 5

2 0

2 5

3 0

3 5

4 0

4 5

5 0

5 5

6 0

6 5

7 0

7 5

8 0

8 5

9 0

9 5

Re

lative

Ab

und

an

ce

1 0 3 4 . 2 5 9 5

1 0 7 5 . 4 1 0 2

1 2 7 7 . 5 7 0 3

1 1 7 6 . 3 6 7 9

1 5 2 0 . 5 3 7 8 1 5 8 5 . 3 8 4 5

8 8 1 . 3 7 6 3

8 2 6 . 4 1 4 7

9 1 7 . 6 2 4 4

1 8 2 8 . 6 1 1 8

1 7 1 4 . 9 3 2 61 3 0 2 . 5 4 9 3

5 1 8 . 3 4 8 99 9 3 . 1 3 9 8

8 0 6 . 7 4 3 8

6 2 0 . 0 5 4 2

5 8 3 . 0 7 8 1 1 9 2 9 . 5 8 2 81 4 7 2 . 9 9 2 7 1 7 4 3 . 2 2 2 97 4 4 . 1 9 5 16 4 0 . 1 7 5 0

4 5 4 . 9 6 7 0 1 8 7 5 . 5 7 6 26 9 8 . 1 8 4 3

1 9 8 6 . 6 3 8 4

NL=1.63E5 Iso-width 5 m/z

54

Figure 2. 3 Representative fragmentation spectra from peptide sequence derived from protein

concanavalin A collected using different microscans

NL = normalization level. Each individual spectrum shown above is well-populated with fragment ions,

and each have similar intensity levels as noted by height of the peaks with similar normalization levels.

Additionally signal strength of fragmentation peaks is well above noise levels.

L C Q _ m i c r o s c a n -2 C -6 -1 4 -0 6 # 3 6 2 7 R T : 9 0 .4 0 A V: 1 N L : 1 .5 2 E 5

T : + c N S I d F u l l m s 2 1 0 5 2 .3 2 @ 3 5 .0 0 [ 2 7 5 .0 0 -2 0 0 0 .0 0 ]

4 0 0 5 0 0 6 0 0 7 0 0 8 0 0 9 0 0 1 0 0 0 1 1 0 0 1 2 0 0 1 3 0 0 1 4 0 0 1 5 0 0 1 6 0 0 1 7 0 0 1 8 0 0 1 9 0 0

m /z

0

5

1 0

1 5

2 0

2 5

3 0

3 5

4 0

4 5

5 0

5 5

6 0

6 5

7 0

7 5

8 0

8 5

9 0

9 5

Re

lative

Ab

un

da

nce

1 5 2 1 .5 7 2 01 1 7 6 .5 8 6 4

1 0 7 5 .7 3 0 5

1 3 4 8 .7 1 8 5

8 8 1 .7 5 9 9 1 4 7 3 .1 3 3 3

9 1 8 .2 7 7 7

1 2 7 7 .5 3 6 1

1 7 1 4 .7 9 1 31 2 4 1 .3 3 6 9

1 9 1 0 .8 5 1 8

8 1 6 .2 6 2 0

9 9 9 .3 1 2 1

9 4 6 .4 2 9 8

1 5 8 5 .9 2 6 06 2 1 .6 7 5 25 1 8 .1 1 0 6

1 4 1 0 .4 3 0 24 5 5 .3 3 5 8 5 8 5 .3 1 7 9 7 5 6 .2 2 4 9

6 5 2 .9 6 9 6 1 8 2 8 .2 1 8 0

6 9 7 .0 7 9 8

1 7 6 2 .8 6 2 3

O p t_ L C Q _ m ic r o s c a n - 3 D # 1 8 7 5 R T : 5 6 .2 4 A V : 1 N L : 3 .0 3 E 6

T : + c N S I d F u l l m s 2 1 0 5 2 .3 2 @ 3 5 .0 0 [ 2 7 5 .0 0 - 2 0 0 0 .0 0 ]

5 0 0 6 0 0 7 0 0 8 0 0 9 0 0 1 0 0 0 1 1 0 0 1 2 0 0 1 3 0 0 1 4 0 0 1 5 0 0 1 6 0 0 1 7 0 0 1 8 0 0 1 9 0 0

m /z

0

5

1 0

1 5

2 0

2 5

3 0

3 5

4 0

4 5

5 0

5 5

6 0

6 5

7 0

7 5

8 0

8 5

9 0

9 5

1 0 0

Re

lativ

e A

bu

nd

ance

1 5 2 0 .6 0 3 38 8 1 .7 1 5 5

1 1 7 6 .3 5 5 2

1 0 7 5 .3 6 3 8

1 0 3 4 .3 0 6 28 2 6 .3 4 7 3

1 7 1 5 .6 0 2 5

1 2 7 7 .5 7 1 8

1 3 0 1 .3 5 7 7

8 0 2 .4 8 1 6

1 5 8 6 .5 5 7 49 2 7 .3 4 1 3

1 4 6 3 .9 1 9 2 1 9 2 8 .6 0 8 2

1 6 4 9 .6 8 4 86 3 1 .2 9 4 2 1 8 7 5 .3 5 0 61 3 4 8 .7 9 1 7

6 9 6 .0 6 9 04 5 5 .1 5 2 8

7 2 8 .3 4 1 85 1 8 .1 8 1 4 1 3 6 8 .6 2 9 6

9 8 1 .5 2 3 95 8 2 .8 0 4 9

1 7 6 2 .7 3 4 6

L C Q _ m ic r o s c a n - 5 - 6 - 1 4 - 0 5 # 2 3 2 8 R T : 9 1 .3 8 A V : 1 N L : 1 .5 2 E 5

T : + c N S I d F u l l m s 2 1 0 5 2 .2 8 @ 3 5 .0 0 [ 2 7 5 .0 0 - 2 0 0 0 .0 0 ]

5 0 0 6 0 0 7 0 0 8 0 0 9 0 0 1 0 0 0 1 1 0 0 1 2 0 0 1 3 0 0 1 4 0 0 1 5 0 0 1 6 0 0 1 7 0 0 1 8 0 0 1 9 0 0

m /z

0

5

1 0

1 5

2 0

2 5

3 0

3 5

4 0

4 5

5 0

5 5

6 0

6 5

7 0

7 5

8 0

8 5

9 0

9 5

1 0 0

Re

lative A

bun

da

nce

1 0 3 4 .3 1 1 0

1 0 7 5 .3 9 9 9

8 8 1 .8 3 4 0

1 2 7 7 .6 4 4 0

1 1 7 6 .3 2 3 2

1 5 2 0 .6 0 6 9

1 5 8 6 .1 4 6 59 1 7 .3 1 7 7

8 0 2 .4 7 5 0

7 6 8 .1 7 5 3 1 3 4 8 .5 5 2 0

4 5 4 .9 6 2 0

1 4 6 3 .5 7 3 06 3 1 .3 5 7 9

1 6 4 9 .4 8 0 7

1 9 1 1 .4 9 8 51 8 2 7 .6 6 2 81 4 1 4 .8 9 9 77 3 6 .8 3 7 0 1 7 1 5 .6 3 6 7

5 1 8 .1 0 7 21 7 6 3 .9 7 7 57 0 1 .5 7 0 3

NL=3.03E6

NL=1.52E5

NL=1.52E5

2 microscans

3 microscans

5 microscans

55

Figure 2. 4 ROC curves of LCQ isolation width EPSM data

Data from 6 replicates of each EPSM isolation width window experiment was pooled and the number of

false identifications (x-axis) versus the number of true identifications (y-axis) was plotted for peptides

detected in each charge state. A) Data from peptides of +1 charge state, B) Data from peptides of +2 charge

state, and C) Data from peptides of +3 charge state. The solid line represents an isolation window of 2 m/z

units, dotted line represents 3 m/z units, dashed line is 4 m/z units, and heavy dashed line is 5 m/z units.

The angled line represents a 95% confidence threshold for each plot. The point at which the straight

dashed line crosses the curve represents threshold levels at which identifications are made with 95%

confidence. From the above plots, it appears that an isolation window width of 2 m/z is best for + 2 and +3

peptides, while an isolation width of 3 m/z units is best for +1 peptides. Isolation widths of 4 or 5 m/z units

do not appear to be significantly different for any charge state.

B) Peptides with +2 charge

C) Peptides with +3 charge

A) Peptides with +1 charge

56

an isolation window of 3 m/z units gives a greater number of true peptide identifications

for the same number of false peptide identifications (Figure 2.4). An isolation width

window of 5 m/z units appears to give the lowest number of true positive identifications

for all charge states. Taken together the ROC curves suggest that an isolation window

width of 2 or 3 m/z units is best for data collection on the LCQ instrument, while an

isolation width of 5 m/z units can possibly be detrimental to results obtained on the LCQ.

However, statistical analysis of average peptide identifications from 6 individual

replicates of isolation width m/z window data filtered using threshold values that returned

a 95% confidence level did not support the suggested optimal isolation window widths

derived from the ROC curves. One-way ANOVA analysis returned no significant

difference between the average total number of identifications obtained (p = 0.9916).

Percentages of peptides identified were computed by dividing the number of

identifications made by the total number of spectra collected, giving a measure of the

efficiency of the search algorithm in assigning sequence identification to the raw spectra.

The average percentage of peptides identified per total spectra collected also showed no

significant difference (p = 0.9054).

Since the EPSM samples are not entirely reflective of real-world total proteome

samples, the possibility remains that significant differences can be discovered in

identification of spectra collected from total proteome samples. For this reason, we

elected to test isolation window widths of 2, 3, and 5 m/z using a total proteome sample.

An isolation window width of 5 m/z yields a greater number of peptide and protein

identifications (Table 2.4), but not significantly more than an isolation window width of

57

Table 2. 4 numbers of protein and peptide identifications from total proteome spectra collected using

differing isolation widths

LCQ

Protein Identifications

Peptide identifications

Isolation

window width

Average Standard

Deviation

Average Standard

Deviation

2 m/z 336 47 1613 326

3 m/z 401 33 2089 186

5 m/z 502 39 2927 281

LTQ

Protein Identifications

Peptide identifications

Isolation

window width

Average Standard

Deviation

Average Standard

Deviation

2 m/z 1084 14 8020 352

3 m/z 1239 85 9684 939

5 m/z 1176 36 8446 756

58

3 m/z. In comparison, an isolation window width of 2 m/z yields a significantly lower

number of peptide and protein identifications than windows of either 3 m/z or 5 m/z (p =

0.0433 for proteins, p = 0.0213 for peptides). Taken together, the above data suggest that

an isolation window width of 3 m/z or 5 m/z will yield a higher number of identifications

than an isolation window width of 2 m/z for an LCQ mass spectrometer.

LTQ linear ion trapping quadrupole mass spectrometer

In contrast to the LCQ data results, examination of the ROC curves for the pooled

EPSM LTQ isolation width data (Figure 2.5) suggests that an isolation window width of

2 m/z yields a lower number of positive peptide identifications for peptides of all charge

states. However, isolation window widths of 3, 4 or 5 m/z units appear to be equivalent

in the number of positive peptide identifications. As visualized in the ROC curves, no

significant differences emerged in mean percentage identifications (p = 0.6432)

or the mean number of true identifications (p=0.2436) in the EPSM isolation width

studies for the LTQ. When further studies were done on proteome samples in the LTQ,

no significant differences emerge for any isolation width tested (Table 2.4) for either

protein identifications (p = 0.1103) or peptide identifications (p = 0.1608).

Effect of varying tandem mass spectrometry (MS/MS) microscans

LCQ quadrupole 3D ion trapping mass spectrometer results

For the LCQ, ROC curves indicate that the use of two or three microscans does

not seem to yield a significantly different number of true positive peptide identifications

(Figure 2.6). However, a higher number of microscans appears to negatively impact the

59

Figure 2. 5 ROC plots of LTQ isolation width EPSM data

Data from 6 replicates of each LTQ EPSM isolation width window experiment was pooled and the number

of false identifications (x-axis) versus the number of true identifications (y-axis) was plotted for peptides

detected in each charge state. A) Data from peptides of +1 charge state, B) Data from peptides of +2 charge

state, and C) Data from peptides of +3 charge state. The solid line represents an isolation window of 2 m/z

units, dotted line represents 3 m/z units, dashed line is 4 m/z units, and heavy dashed line is 5 m/z units.

The angled line represents a 95% confidence threshold for each plot. The point at which the straight

dashed line crosses the curve represents threshold levels at which identifications are made with 95%

confidence. From the above plots, it appears that an isolation window width of 2 m/z negatively impacts

the total number of identifications that can be made, regardless of charge state, while an isolation width of

3, 4, or 5 m/z units do not appear to be significantly different for any charge state.

B) Peptides with +2 charge

C) Peptides with +3 charge

A) Peptides with +1 charge

60

Figure 2. 6 ROC curves of pooled LCQ EPSM microscan data

Data from 6 replicates of each EPSM microscan experiment was pooled and the number of false

identifications (x-axis) versus the number of true identifications (y-axis) was plotted for peptides detected

in each charge state. A) Data from peptides of +1 charge state, B) Data from peptides of +2 charge state,

and C) Data from peptides of +3 charge state. The solid line represents 2 MS/MS microscans, dotted line

represents 3 MS/MS microscans and the dashed line represents 5 MS/MS microscans. The angled line

represents a 95% confidence threshold for each plot. From the above plots, it appears that 5 MS/MS

microscans has a negative effect on the number of identifications that can be made, while 2 or 3 microscans

give the maximal number of identifications.

B) Peptides with +2 charge

C) Peptides with +3 charge

A) Peptides with +1 charge

61

number of peptides identifications obtained from the protein standard mix. The statistical

analysis of the microscan data, however, tells a different story. A significantly lower

mean number of average percentage identifications was found for 2 microscans (p =

0.0078) while there was no significant difference between mean number of identifications

of 3 and 5 microscans. The average number of true identifications also shows a

significant difference between the mean numbers of identifications obtained using

different microscans, with 5 microscans giving significantly lower number of IDs than

either 2 or 3 microscans. Taken together this data suggests that 2 MS/MS microscans

negatively impacts the ability of the search algorithm to identify peptides from raw

spectra. On the other hand, 5 MS/MS microscans does not affect the ability of the search

algorithm to identify peptides from raw spectra, but instead results in a lower number of

identifications due to the increased amount of time required to collect the spectra. From

this data, 3 MS/MS microscans appear to be the optimal number of microscans for the

LCQ.

Further testing was done using 3 and 5 MS/MS microscans for the proteome data

in order to determine if the increased amount of time to collect the scans would impact

the total number of protein identifications made. Average peptide and protein counts are

presented in Table 2.5, and box plots of LCQ proteome data in Figure 2.7. Although

there is a slight decrease in the number of protein and peptide identifications with 5

MS/MS microscans versus 3 MS/MS microscans, this difference is not statistically

significant for peptides (p = 0.5711) or for protein identifications (p = 0.2046).

62

Table 2. 5 Average numbers of protein and peptide identifications from total proteome spectra

collected using differing MS/MS microscan settings

LCQ

Protein Identifications

Peptide identifications

No. of MS/MS

Microscans

Average Standard

Deviation

Average Standard

Deviation

3 404 33 2089 186

5 370 16 2042 263

LTQ

Protein Identifications

Peptide identifications

No. of MS/MS

Microscans

Average Standard

Deviation

Average Standard

Deviation

1 1237 78 9638 774

2 1239 85 9684 939

63

Figure 2. 7 Box plot representation of statistical results from pooled LCQ proteome microscan data

Data from 6 individual replicates was analyzed via one-way ANOVA and means and standard deviation

determined. The x-axis represents the number of MS/MS microscans (either 3 or 5) tested on LCQ

proteome samples. The box encloses two standard deviations within the data range, while the star in the

center represents the mean, and the line represents the median. A) Mean numbers and standard deviation of

total numbers of protein identifications. B) Mean numbers and standard deviation of total numbers of

peptide identifications. Statistical analysis showed that there is no significant difference between the

groups in the numbers of protein (p = 0.2046) or peptide (p = 0.5711) identifications obtained when

proteome samples are tested.

Si de-by-Si de Boxpl ot s t o compare Prot ei n I Ds

3 5

425

450

475

500

525

550

P

rotID

Mi croscan

Si de-by-Si de Boxpl ot s t o compare Mi croscan Prot eome Pept i de I Ds

3 5

2400

2600

2800

3000

3200

PepID

Mi croscan

A) Protein identifications

B) Peptide identifications

64

LTQ linear ion trapping quadrupole mass spectrometer

The LTQ microscan data showed similar results to those for the LCQ.

Examination of ROC curves derived from EPSM samples run on the LTQ (Figure 2.8)

shows that 2 microscans gives a noticeably higher number of true positive peptide

identifications for peptides in all charge states, although the difference is less pronounced

for peptides with a charge state of +3. Again, the number of microscans showed

significant differences in average percentage identifications (p < 0.0001). But here, all

three average percentage IDS for each microscan are significantly different from one

another. There are also significant differences in the numbers of peptide identifications.

The average number of identifications from 2 microscans overlaps with both 1 and 3

microscans. There is no statistically significant difference between the number of

identifications found in data collected using 2 microscans and that found using 3

microscans. Nor is there a significant difference between the numbers of identifications

found using 2 microscans versus 1 microscan. However, 3 microscans give significantly

lower numbers of identifications than 1 microscan (p = 0.0024).

Due to the lower number of peptide identifications obtained with 3 MS/MS

microscans, this experimental condition was not included in proteome studies.

Proteome data, given in Table 2.5, shows very little difference between average numbers

of proteins and peptides identified from data collected with 1 MS/MS microscan and that

collected with 2 MS/MS microscans. For a more complex soluble proteome sample, no

significant difference in mean number of protein (p = 0.9061) or peptide (p=0.9858)

identifications are found between 1 and 2 microscans (Figure 2.9), suggesting that

65

Figure 2. 8 ROC curves representing LTQ pooled EPSM microscan data

A) Number of EPSM peptides identified of +1 charge state, B) Number of +2 charge state EPSM peptides

identified, and C) Number of +3 charge state EPSM peptides identified. Solid line represents data collected

using 1 MS/MS microscan, dotted line represents data collected with 2 MS/MS microscans, and dashed line

with 3 MS/MS microscans. From the above plots, 2 microscans appears to be the clear winner for all

charge states.

B) Peptides with +2 charge

A) Peptides with +1 charge

C) Peptides with +3 charge

66

Figure 2. 9 Box plots representing statistical analysis of LTQ microscan proteome data

Data from 6 individual replicates was analyzed via one-way ANOVA and means and standard deviation

determined. The x-axis represents the number of MS/MS microscans (1 or 2 microscans) tested on LTQ

proteome samples. The box encloses two standard deviations of the data range, while the star in the center

represents the mean, and the line represents the median. A) Mean numbers and standard deviation of total

numbers of protein identifications. B) Mean numbers and standard deviation of total numbers of peptide

identifications. As with the LCQ data, statistical analysis showed that there is no significant difference

between the groups in the numbers of protein (p = 0.9061) or peptide (p = 0.9858) identifications obtained

when proteome samples are tested.

Si de-by-Si de Boxpl ot s t o compare Prot ei n I Ds

1 2

1100

1150

1200

1250

1300

1350

ProtID

Mi croscan

Si de-by-Si de Boxpl ot s t o compare Mi croscan Prot eome Pept i de I Ds

1 2

8000

8500

9000

9500

10000

10500

11000

Pep

ID

Mi croscan

B) Peptide identifications

67

SEQUEST is not significantly affected by improvements in spectral appearance derived

from spectral averaging of LTQ scans.

Discussion

The advantage to using a small set of known proteins for initial analysis is that it

provides a method for establishing true positive identifications from false positive

identifications. Initial searches against a database containing both forward and reversed

protein sequences from each protein included in the EPSM mix allowed for determination

of specific data filter levels that would ensure a 95% false positive rate. It also allowed

for determination of numbers of true and false identifications for each threshold level,

used in creating the ROC curves discussed below. The disadvantage to using a small set

of known proteins mixed at equimolar concentrations is that this mixture does not

represent a real world sample. Although it can narrow the window of test conditions, an

EPSM can not necessarily give an accurate indication of results in extremely complex

samples where protein concentrations vary across a wide range. For this reason two levels

of tests were done, with the second level including investigation of total protein

identifications obtained from total soluble proteome preparation of R. palustris.

In order to determine appropriate test levels for proteome experiments, Receiver

Operating Charateristic (ROC) curves were compiled from initial protein standard data.

Our ROC curves were constructed by plotting the number of true positive peptide

identifications versus the number of false positive identifications at a given threshold

level. They were initially used to define the performance of radio receivers in terms of

signal-to-noise ratios, and are now commonly used to ascertain the efficacy of medical

68

tests [127]. They have also been used in comparisons of performance of search

algorithms when searching spectra derived from known protein standard mixes [7]. The

advantage of using a protein standard mix for creating ROC curves is the presence of a

known set of proteins. Unlike medical tests where the detection of a true positive test

result can never truly be “known,” a true identification of a peptide from a defined set of

proteins has a greater chance of being a true positive, while false identifications can

easily be determined.

When constructing ROC curves from EPSM data, the data was pooled to provide

a greater statistical representation of peptide identification numbers. On the ROC plot,

the vertical axis is number of true positive identifications, while the horizontal axis is the

number of false positive identifications represented by the number of peptide

identifications made to proteins known not to be included in the EPSM mixture (Figure

2.10). The curves are constructed by determining the number of true positives and the

number of false positives at a number of different threshold levels. Identification

numbers are separated by charge states of the parent ion because tandem mass spectra

fragment ions derived from parent ions of different charge states are identified with

different levels of efficiency. Each point on each line of the ROC curves represents the

number of true positives versus the number of false positives for a given threshold level.

Ideally, the curve would be a straight ninety-degree angle, where 100% of the peptides

present would be identified positively before any false positives are identified.

Comparisons between curves are made by observing which curve is on top of another,

with the curve on top indicating a higher number of overall peptide identifications [7].

69

Figure 2. 10 Development of an ROC curve (Adapted from http://www.anaesthetist.com/mnm

/stats/roc/)

True negative identifications (TNF, dark blue curve) and true positive identifications TPF, red curve) are

represented by Gaussian distribution curves on the left side of the figure above. In the above illustration,

the blue curve represents the total number of false negatives while the red curve represents the total

numbers of true positive identifications in a given data set. Light blue represents those identifications which

are incorrectly identified as true (FPF, false positive fraction). Pink represents those true identifications

incorrectly classified as false (FNF, false negative fraction). Overlap of the two curves is indicative of the

ability to discriminate between true and false identifications. The green line represents a given threshold.

ROC curves on the left are derived from moving the threshold level (green line) and plotting a point

representing an x-axis value of false positives and a y-axis value of true positives for that given threshold

level.

70

For this study the curves were not normalized due to technical difficulties in changing the

R script from which they were derived. A second level of statistical analysis was then

done for each experimental condition at one specific threshold level of 95% false positive

rate to determine if results at that threshold level were statistically significant. The

isolation m/z width and the number of microscans tested using a proteome sample were

based on a combination of results obtained from analysis of pooled EPSM data sets as

represented by the ROC curves and from further statistical analysis of individual EPSM

sample data sets.

Tandem mass spectrometry is performed through isolating and fragmenting a

selected ion of a desired mass-to-charge ratio (m/z). Radio frequency (RF) voltage is

applied to isolate the parent ion, but there are a number of factors to consider in that

isolation. The isolation window width should be chosen such that the entire isotopic

packet of the ion of interest is isolated while at the same time precluding isolation of a

co-eluting ion of a similar m/z. If the isolation window width is too small, the entire

isotopic packet of the ion of interest may not be collected, which may in turn significantly

affect the ability to identify the isolated peptide. On the other hand, if the isolation

window width is too large, the risk of selecting a co-eluting peptide of similar m/z

increases. Co-eluting peptides could lead to the presence of additional fragment ions

and/or an increase in noise levels, and could thus result in an inability to correctly

identify the peptide sequence of the parent ion of interest. Choice of an isolation

window of 2 m/z remains common because this is the default isolation window in the

XCaliber software, while isolation window widths of 3 m/z are common in literature. In

contrast, isolation widths of 5m/z have been used in the OBMS group for proteome

71

analysis, with greater identification numbers reported for this larger m/z isolation

window. We therefore chose a range of isolation window from 2 m/z units to 5 m/z units

for our initial EPSM tests in order to determine the effect of varying the width of this

window. Our results from the set of isolation window width tests indicated that although

the simpler EPSM mixture appeared to have significant differences in ROC curves

(figures 2.4 and 2.5 for the LCQ and LTQ, respectively), statistical analysis did not

support that conclusion. More complex proteome mixtures, however, did show a

significant decrease in the numbers of identifications obtained with an isolation window

width of 2 m/z for the LCQ (Table 2.4) but no significant differences in identifications

for any isolation width in the LTQ (Table 2.4). A narrow isolation window width may

lead to a lesser ability to isolate parent ions in the LCQ, especially if other isolation

parameters are not functioning optimally, due to the lower amount of ions in the trap than

in the LTQ linear ion trapping quadrupole instrument.

When considering the effect of varying the number of averaged scans designated

to give a final spectrum, both the quality of the acquired final spectrum and the amount of

time required to acquire it must be taken into consideration. Mass spectra are obtained by

a scan function consisting of an ion injection time, fixed time length prescan followed by

an additional variable ion injection time, set by the ion density found in the prescan to

result in a specified number of ions in the trap. The final observed spectrum is a result of

averaging a number of microscans (Figure 2.1). Higher numbers of microscans result in

a spectrum with more normalized peaks, greater signal-to-noise ratio and a greater

number of fragment ion peaks, but also increase the amount of time needed to perform

each scan. Although the appearance of the spectra may be cleaner with more

72

microscans, the appearance may not make a difference when it comes to identifying the

peptide which gave rise to it. In fact, when examining spectra from the identical peptide

collected using different experimental conditions (Figure 2.3), no discernible difference

in spectral quality is noted.

The additional time required for more microscans may instead hinder peptide

identifications because less time is available to collect spectra during each point in the

chromatographic run. For instance, if you have a scan function lasting for a time period

of 125 msec, and you are performing 4 microscans per observed spectrum, you have a

total spectral collection time of 500 msec per spectrum, thus limiting the number of

spectra you can collect to two spectra per second. This longer scan time may in turn

decrease the number of identifications that can be made. We found that increasing the

time required for doing additional microscans did have a negative impact on the number

of identifications acquired, with a far greater difference noted in data obtained on the

LTQ mass spectrometer than the LCQ mass spectrometer.

But the question remains, what number of microscans can produce the happy

medium of adequate spectra plus maximum identifications? Does an increased number

of microscans increase the search algorithm’s ability to distinguish true and false

identifications? Earlier studies have yielded conflicting results when attempting to

answer these questions on either a 3-dimensional (3D) trapping quadrupole or a linear

trapping quadrupole instrument. When varying only the number of microscans while

leaving all other mass spectrometry parameters constant, the number of microscans does

seem to result in significant differences in the number of identifications [60] or in better

73

scores for each identified peptide [59]. However, when varying a number of parameters

simultaneously and considering interactions between these parameters in a multi-

parametric statistical modeling approach [34, 57, 58], the number of microscans in a 3D

trapping quadrupole instrument does not emerge as a significant factor in the number of

identifications acquired [57, 58]. In our tests, the number of microscans was held

constant for the parent ion, and settings were based on the standards used in the industry

(typically 3 microscans for a full scan in LCQ, and 1 microscan for full scans in the

LTQ). The number of microscans used to collect the tandem (MS/MS) mass spectra

were varied according to experimental parameters shown in Table 2.3. Further,

microscan numbers for the LCQ quadrupole ion trap were larger than those tested for the

LTQ linear ion trap due to the larger number of ions that can be trapped within a linear

trap. Performing a higher number of microscans while acquiring spectra in the LCQ ion

trap can help to compensate for the lower ion population in the trap. The LTQ ion trap

has a much higher ion population, resulting in better ion statistics, and thus a lower

number of microscans should be sufficient for obtaining pretty spectra as well as

maximum identifications [56]. Since only 1 microscan should be required, more spectra

can be collected due to the decreased amount of time required to collect spectra. Just as

was done for the isolation width data, initial tests for the microscan data were performed

using a known set of proteins mixed in equimolar amounts, and results used in testing a

more complex proteome mixture.

Analysis of microscan data indicated that significant differences emerged in data

collected using different numbers of MS/MS microscans in the LCQ (Figure 2.7). Even

though 5 microscans gave the lowest numbers of identifications (Table 2.5), statistical

74

analysis did not show a significant difference in number or percentage of identifications

for the LCQ microscan data. However, 2 microscans resulted in a significantly lower

percentage of spectra identified for the total spectra collected, suggesting that a lower

number of MS/MS microscans negatively impacts the ability of search algorithms to

assign identifications to the total numbers of spectra being collected. In other words,

even though there are more spectra being collected due to the shorter collection time, less

of these spectra are being identified. A different picture emerged for the LTQ microscan

data, however, with significant differences noted between all experimental conditions for

the EPSM mixture (Figure 2.9). Performing only 1 MS/MS microscan when acquiring

data on the LTQ did result in higher numbers of identifications (Table 2.5), most likely

due to the lesser amount of time required to collect the spectra, but the difference

between higher numbers obtained with 1 microscan versus the numbers obtained with 2

microscans proved to be statistically insignificant.

Conclusions

Isolation width window studies showed no significant differences between

experimental conditions tested for either protein standard mix or proteome samples. A

smaller m/z isolation window slightly increased the number of identifications in the LCQ,

while a slight decrease was noted in data from the LTQ for smaller m/z isolation

windows. Opening the width of the isolation window to 5 m/z units did not change the

number of identifications obtained in this study for either LCQ or LTQ data. The

decreased probability of identifications due to co-eluting peptides was not seen in this

study. However, determination of interference in the MS/MS from fragmentation of both

the parent ion of interest and a co-eluting peptide with a larger m/z window may only be

75

noticed with manual processing of individual spectra. It is possible that automated

processing with the available software simply can not distinguish the presence of a co-

eluting peptide, perhaps making a correct identification regardless of the presence of

additional fragment ions or an increase in signal-to-noise ratio. It is also possible that a

false positive identification is made in this instance, or that no identification is made at

all. The question of interference of co-eluting peptides with the ability of the search

algorithm to identify them could possibly be answered by testing a less complex mixture

of only one protein, or a mixture of two similar peptides. Since no significant difference

was noted in identifications obtained in isolation width experiments, it is recommended

that the industry standard of an isolation width of 3 m/z units be maintained.

Microscan studies also yielded no significant differences for proteomic studies.

Less microscans on the LCQ are not beneficial for peptide or protein identification,

perhaps due to the limited number of ions in the trap resulting in an MS/MS spectrum

that contains more noise or does not have enough fragment ions present for identification.

In contrast, performing more microscans in an LTQ MS/MS experiment does not

significantly increase the number of peptide (and thus protein) identifications obtained.

The greater amount of time required for additional scans results in less spectra collected.

If the same percentage of spectra are being identified regardless of the number of

microscans, then a lower number of spectra will result in less identifications just due to

simple percentage identified. Like other studies before this one, no significant

differences in number of identifications obtained were noted in the lower number of

microscans tested for the proteome samples. Thus, the number of MS/MS microscans

76

used for the LCQ should be kept at 3, while 1 MS/MS microscan is sufficient for

maximal identifications in the LTQ.

77

Chapter 3. Proteomic Investigations of A. brasilense strains

Sp245 and Sp7 under nitrogen fixing and non-nitrogen fixing

conditions

Introduction

Azospirillum brasilense, introduced in Chapter 1, are free-living soil bacteria with

plant growth promoting capacities, and as such are important organisms in the study of

plant-microbe interactions. The physiological versatility of A. brasilense gives them a

competitive advantage in utilization of the limited nutrients found within the rhizosphere.

Because of the positive effect on the growth of grasses when soil is inoculated with

Azospirillum species, genetic studies have been carried out on Azospirillum species for a

number of years. The genome structure has been found to be highly mobile, containing

a number of phage sequences integrated into the genome of all Azospirillum species

[128]. All species carry a number of plasmids [129, 130]. Earlier studies revealed the

presence of 5 megareplicons, 2 circular and 3 assumed to be linear, within all A.

brasilense strains studied [130], but the number of plasmids are variable between strains

and also between different cultures of the same strain. A plasmid of ~142-kb with a

mass of ~85-MDa from one culture of A. brasilense Sp245 has recently been

characterized, but is not present in the recently sequenced Sp245 strain [131], thus

demonstrating the plasmid variability among different strains or even different cultures of

the same strains.

One variant of A. brasilense strain Sp245 has been sequenced by the Joint

Genome Institute (JGI) and the genome sequence can be found at http://genome.ornl.

gov/microbial/abra/19sep08/. This draft genome sequence is 7.5 Mb in 70 contigs of 20

78

reads or greater, contains 68.4 % GC content, and has 7043 candidate protein-encoding

gene models available. From this draft sequence, a protein sequence database consisting

of 6927 potential protein coding sequence translations has been derived.

A. brasilense strain Sp7, in contrast, contains a smaller genome of ~6.3Mb, with

variability in the plasmid sizes seen between it and the Sp245 genome structure [130].

Additionally, Sp7 has one essential plasmid of 151.3 kb with a mass of 90 kDa, named

pRHICO or p90, which contains genes necessary for exopolysaccharide (cell wall)

biosynthesis, as well as for synthesis and export of polysaccharides. Other genes

involved in beta-lactamase expression for resistance to ampicillin are also present on the

pRHICO plasmid [132, 133]. A number of genes on this plasmid have been shown to be

present within megareplicons of the Sp245 strain. Strain Sp7 is closely related to strain

Sp245, but its genome sequence is not yet available. Searching the NCBI nucleotide

database (http://www.ncbi.nlm.nih.gov) using the term “Azospirillum brasilense Sp7”

returns 67 results of individual genes sequenced from the A. brasilense Sp7 strain.

A BLASTx [134] search comparing the translated coding sequences to those in the Sp245

genome revealed all but 1 Sp7 gene sequence to be present in the Sp245 genome with the

remaining genes having 90-100% similarity to Sp245 sequences at a protein level.

Therefore, in this work we have used the Sp245 database in the analysis of data derived

from the Sp7 species. A small percentage of expressed proteins may go undetected in the

Sp7 strain data by using the Sp245 genome for searches, but the large degree of similarity

in 95% of the translated Sp7 coding sequences to those in the Sp245 genome suggests

79

that those proteins detected are likely to be accurate. Using the same protein databases

for both species also facilitates direct comparisons of protein expression by each strain.

Protein products from a variety of genes are involved in nitrogen fixation. These

genes encode not only the structural components of the nitrogenase itself (discussed in

Chapter 1), but also those genes involved in synthesis of both the nitrogenase and the

iron-molybdate cofactor, in electron transport to the nitrogenase, and in regulation of

nitrogen fixation [96]. The nifHDK operon is about 5500bp and encodes the basic

structural components for the molybdenum nitrogenase. Additional related nif genes can

be found both upstream and downstream of this cluster [135]. Altogether, these genes are

found clustered in a 30 kb DNA region in the genome [98]. The fix genes proposed to be

involved in electron transfer to the nitrogenase components [136-138] are found in close

proximity to nitrogenase structural component genes within the genome. Since

Azospirillum brasilense has served as a model organism for nitrogen fixation among soil

bacteria of genus Azospirillum, nitrogen fixation genes have been well characterized [91].

While Sp245 can enter the intercellular space of root cells and live within this

somewhat protected environment, Sp7 does not possess this ability, but instead forms

clumps on the outer surface of the root hairs [93]. Since these strains are suited to grow

in different environments, their physiological responses to external stimuli are likely to

be slightly different as well. For instance, in response to heavy metal stimuli there are

noticeable differences in physiological effects between strains [92]. Sp7 shows

accumulation of poly-hydroxybutyrate (PHB) compounds within the cell in response to

metal stress, while Sp245 does not. On the other hand, both strains show decreased

production of indole-3-acetic acid (IAA) in response to metal stress, but the decrease is

80

far greater for Sp245 than Sp7 [92]. Thus, comparison of entire proteome expression for

these two strains grown under two different growth conditions should reflect the

differences in physiology required for life in different ecological niches.

This study presents comparative proteomic investigations of 2 different strains of

A. brasilense, Sp245 and Sp7, grown in minimal media under both nitrogen limited

(nitrogen fixing) and nitrogen replete (non nitrogen fixing) conditions, with vanadium

trichloride added to the media of one Sp245 nitrogen fixing culture. The goal of this

research was to characterize the overall physiology of both strains under these defining

growth conditions and to reveal putative differences and similarities. Bottom-up mass

spectrometry employing a MudPIT separation technique [17, 19, 20] coupled to an LTQ

linear quadrupole mass analyzer was used to investigate protein expression under each

growth condition. Comparison of global changes in protein expression between these

two growth conditions were investigated in more detail and indicate that while most of

the proteins detected under both conditions are those involved in protein translation,

ribosomal syntheses and biogenesis, nitrogen fixing cells expressed more proteins

involved in nutrient synthesis and manipulation. In addition to adding information to

complete genome sequence annotation, analysis of changes in the physiology of A.

brasilense cells in presence or absence of combined nitrogen by proteomics provides

further insight into how this diazotroph adapts to nitrogen-fixing conditions.

81

Materials and Methods

Cell growth

Overnight starter cultures (5 mL) were inoculated from fresh plates. Starter

cultures were grown overnight at 27°C in a shaking water bath in minimal media

containing malate as carbon source and ammonium chloride as nitrogen source. Cells

were pelleted from starter cultures and washed with appropriate growth media. Base

media for all cultures was minimal media (MMAB) [139] with 20 mM malate as carbon

source, and ammonium chloride as nitrogen source where appropriate. All Sp7 strain

cultures and Sp245 strain control (non nitrogen fixing) cultures were supplemented with

molybdate, while one Sp245 nitrogen fixing culture was supplemented with molybdate

only and a second culture supplemented with vanadium trichloride only. Starter cultures

were resuspended with appropriate media and used to inoculate 500 ml cultures for

control (non-nitrogen fixing) growth, 250 mL cultures for nitrogen fixing growth.

Nitrogen fixation requires a great deal of energy and continuous optimal oxygen

concentrations, so growth of nitrogen fixing cells is slower than those growing in

nitrogen sufficient conditions. Cells grown under nitrogen fixing conditions exhibit a

doubling time of 170 minutes while control (non nitrogen fixing) cells have a doubling

time of 120 minutes [140]. Further, OD of cells grown under nitrogen fixing cultures

never reaches high levels, tending to level off at or below an OD600 of 0.2 – 0.3 [140].

Therefore, each growth condition was optimized as follows: nitrogen fixing cultures were

grown at 25°C without shaking to early log phase (OD600 = 0.1 - 0.2) to minimize

exposure to high levels of oxygen Azospirillum spp. are microaerophilic diazotrophs, and

as such, control cultures (non nitrogen fixing) were grown under optimum growth

82

conditions (shaking and in presence of ammonium) at 25°C on an orbital shaker to mid-

log phase (OD600 = 0.5 – 0.6). Cells were harvested by centrifugation at 8000 rpm for 10

minutes, washed twice with 50 mM Tris (pH 7.9), then pelleted by centrifugation at 8000

rpm for 10 minutes. Two separate cultures were grown under each condition, here

termed biological replicates. Cell pellets from these two biological replicates were

pooled for subsequent proteome preparation in order to ensure sufficient protein

concentration allowing technical replicates (two mass spectrometry runs of the same

sample) to be performed. Wet cell pellet weight was recorded and cell pellets were

frozen for later analysis.

Proteome preparation for LC/LC-MS/MS

Frozen cell pellets (0.1 g for each sample) were resuspended at a rate of 500 µl

lysis buffer/ 0.1 g wet cell pellet weight in lysis buffer of 6 M guanidine hydrochloride,

10 mM DTT solubilized in 50 mM Tris-HCl, 10 mM CaCl2. Resuspended cells were

then further lysed by sonication. Lysate was centrifuged at 18,000 xg for 20 minutes to

clear cellular debris. Supernatant was collected, with 1 ml of lysate used for immediate

tryptic digestion and the rest frozen at -80°C for later digestion. Protein concentration

was not quantified due to interference of guanidine salt with commonly used protein

quantification assays. Trypsin digestion was performed according to manufacturer’s

recommendations. Briefly, 10 mM DTT was added and lysate was incubated at 60°C for

1 hour. Lysate was then diluted 6-fold with trypsin digestion buffer (50 mM Tris-HCl,

10 mM CaCl2, 10 mM DTT, pH 7.9) and 20 µg sequencing-grade trypsin (Promega,

Madison, WI) was added to each sample. Samples were incubated overnight at 37°C

with gentle rotation. An additional 20 µg of trypsin was added the following morning

83

and samples were subsequently incubated for an additional 5-6 hours at 37°C with gentle

rotation. Digestion was halted by addition of 5 µl formic acid to the 5 ml lysate.

Samples were then desalted using Sep-Pak Plus C-18 solid phase extraction (Waters,

Milford, MA) following manufacturer’s recommendations, and subsequently

concentrated and solvent-exchanged into 100% HPLC-grade H2O, 0.1% formic acid

using vacuum centrifugation (Savant, Thermo Scientific). Samples were aliquoted into

40 µL volumes and stored at -80°C until analysis.

LC/LC-MS/MS analysis

Instrument performance was evaluated before beginning experimental mass

spectrometry proteomic runs through first running an R. palustris soluble proteome

standard. Proteome standards were run according to the protocol described in Chapter 2,

with a 5-step MudPIT setup [17, 19]. Parameters evaluated from the standard included

chromatographic separation capacity, normalization levels of the mass spectra, evaluation

of signal-to-noise ratio and numbers of proteins resulting from data searches as described

in Chapter2. An additional R. palustris proteome standard was run at the end of

experimental runs and results compared in order to ensure that the LTQ mass

spectrometer was functioning optimally throughout the experimental runs.

Proteome samples were analyzed in quadruplicates via Multi-dimensional Protein

Identification Technology (MudPIT) [17, 19, 20] with triphasic columns. Columns were

individually packed using a pressure cell (New Objective, Woburn, MA). Back columns

were loaded first with 3 cm of Luna 5µm strong cation exchange (SCX) resin

(Phenomenex, Torrance, CA) followed by 3 cm of Aqua 5 µm C-18 reverse phase resin

84

(Phenomenex). Proteome aliquots (50 µl) were loaded directly onto the back column via

pressure cell and subsequently coupled to the front column. Final protein concentration

loaded onto the back column was evaluated by the intensity level of the chromatogram,

and amount of protein loaded was adjusted in subsequent runs to achieve approximately

equal loading amounts. Front columns were pulled to a tip with an inside diameter of 5

µm using a P-2000 laser puller (Sutter Instruments, Novato, CA), and packed with 17 cm

of Aqua 5 µm C-18 reverse phase resin. This column acts as the resolving column for

peptides eluted from the back column. For analysis, the combined columns were placed

directly in-line with the linear trapping quadrupole (LTQ) mass spectrometer

(ThermoFinnigan, San Jose, CA) using a Proxeon source.

Chromatographic separation was accomplished with an Ultimate HPLC system

(LC Packings, a division of Dionex, San Francisco, CA) providing a flow rate of 100 µl/

minute which was split prior to the resolving column such that the final flow rate at the

tip was ~300 nl/minute. Twelve two-dimensional (2D) chromatographic steps were

done. An initial 1 hour gradient from buffer A (95% water, 5% acetonitrile, 0.1% FA) to

buffer B (70% acetonitrile, 0.1% FA) bumped the peptides from the initial reverse phase

column onto the strong cation exchange column. Subsequent cycles included 2 minute

salt pulses with varying percentages of 500 mM ammonium acetate (10, 15, 20, 25, 30

35, 40, 45, 50, 60%) to first elute subsets of peptides according to charge, followed by a 2

hour gradient from buffer A to buffer B, to further separate peptides by hydrophobicity.

The final chromatographic step consisted of a 20 minute salt pulse of 100% 500 mM

ammonium acetate, followed by a 2 hour A-to-B gradient.

85

Data collection was controlled by Xcaliber software (ThermoFinnigan). Data was

collected in data-dependent mode using parameters established in Chapter 2 with one full

scan followed by 6 dependent scans, each with 2 microscans. Dynamic exclusion was

employed with a repeat count of 1, repeat duration of 60 and exclusion list size of 300

and duration of 180 msec. Isolation mass width was set at 3 m/z units.

Data Analysis

The two best technical replicates were used in all data analysis. An Azospirillum

protein database based on the newly sequenced Sp245 genome was constructed from

CDS text files based on translations of the coding sequences called in the draft genome

sequence (http://genome.ornl.gov/microbial/abra/19sep08/). A list of common

contaminants was appended to the gene call sequences, and all coding sequences,

including contaminant sequences, were reversed and appended to the forward sequences

in order to serve as distractors. The protein database can be downloaded at

https://compbio.ornl.gov/mspipeline/azospirillum. Due to the high degree of similarity of

translated coding sequences available for the Sp7 strain to the translated coding

sequences of the Sp245 strain, other databases were not appended for the Sp7 database

searches. Further, additional databases were not included in the search due to the large

increase in time required for searches of larger databases. All MS/MS spectra for both

Sp245 and Sp7 strains were searched against this database using SEQUEST [3]

specifying tryptic digestion, peptide mass tolerance of 3 m/z and a fragment ion tolerance

of 0.5 m/z. Additionally, search parameters included two dynamic modifications: 1.

methylation represented by a mass shift of +14 m/z on aspartate residues, and 2.

deamidation followed by methylation represented by a mass shift of +15 m/z on

86

glutamine residues. Output data files were sorted and filtered with DTASelect [54] ,

specifying XCorr filter levels of 1.8 for peptides with a charge state of +1, 2.5 for those

with charge state +2 and 3.5 for charge state +3, minimum delta CN of 0.08, semi-tryptic

status and 2 peptides per protein identification. DTASelect files in both html and text

formats are posted at https://compbio.ornl.gov/mspipeline/azospirillum/study1/

status.html.

Results were imported into Microsoft Access for data analysis. Results for each

technical replicate run of an individual growth state were combined to give an average

number for that growth state, and combined results were used in all analysis. From the

number of identifications in the reverse direction, peptide false positive rates were

determined using the formula %FP = 2[No. reverse ID/ (no. reverse ID + no. real ID)]

[18]. In order to determine relative abundance of a given protein in a sample, normalized

spectral abundance factors (NSAF) were calculated for each individual protein k using

the formula NSAFk = (SpC/L)k / Σ (SpC/L)n, where SpC is the total spectral count for all

peptides contributing to protein k, L is the length of protein k, and n is the total number of

proteins detected in the sample [72]. For those proteins found in both replicate runs, an

average of two NSAF values was calculated. Pairwise comparisons of the calculated

NSAF for each protein in each growth state were used to determine up- and down-

regulation. Experiments done by Zybailov et al [24] indicated that a 1.4-fold increase in

NSAF was significant for their yeast membrane proteome data set with 9 replicate

analyses. With only two replicates, statistical significance is difficult to determine, so in

order to ensure significance of up-regulation in this data set, data was filtered such that

only a 5-fold or greater difference in the calculated NSAF was reported.

87

For analysis of fold-change of proteins between growth states, proteins were

required to be present in both growth conditions, so the NSAF values could be compared

directly across growth conditions. Because NSAF values are reflective of the abundance

of a protein within a sample (discussed in chapter 1 of this work), comparison of the

NSAF level of a protein in nitrogen fixing conditions with that of the same protein in non

nitrogen fixing conditions gives an estimate of the degree of up-regulation or down-

regulation of that protein when conditions change. Dividing the NSAF value of each

single protein found in nitrogen fixing conditions by the NSAF value of that same protein

in non nitrogen fixing control conditions gives an approximate fold-change value of that

protein. Proteins present in only one growth state were not considered in this analysis

because of the possibility that these proteins are expressed only in one growth state and

not the other, making it impossible to determine if levels of up-regulation or down-

regulation are appropriate.

Results

Overall proteome evaluation

The newly sequenced Azospirillum brasilense Sp245 genome contains 6927

candidate gene coding sequences. As shown in Table 3.1, our proteomic study identified

a total of 1289 proteins (19%) in 2 combined technical replicates of Sp245 cultures

grown under optimal growth conditions (in presence of ammonium and with shaking),

herein termed control conditions and 1372 (20%) and 1269 (18%) proteins in Sp245

cultures grown under nitrogen fixing conditions with molybdate and vanadium,

respectively. Reproducibility between technical replicate runs was 67% for both Sp245

control cultures and nitrogen fixing with molybdate, while reproducibility between

88

Table 3. 1Total numbers of protein and peptide identifications in each sample

Sample Number of Protein Identifications Number of Peptide Identifications

Replicate 1 Replicate 2 Combined Dataset Replicate 1 Replicate 2

Sp245 Control 976 1177 1289 7220 10259

Sp245 Nitrogen Fixing - Mo 1155 1159 1372 9133 8324

Sp245 Nitrogen Fixing - V 1067 1171 1269 9220 6167

Sp7 Control 812 851 1117 6169 7540

Sp7 Nitrogen fixing 985 1017 1181 8076 7839

89

nitrogen fixing vanadium culture technical replicates was slightly higher at 71%. False

positive rates at a peptide level for Sp245 cultures ranged from 1.4% for non nitrogen

fixing optimally-growing cultures to 2.8% for nitrogen fixing cultures with molybdate.

Sp7 cultures yielded slightly lower totals of 1117 (16%) and 1181 (17%) protein

identifications for data from combined technical replicates of control optimal growth

(non-nitrogen fixing) cultures and nitrogen fixing cultures, respectively. Optimal growth

controls and nitrogen fixing Sp7 cultures had 77% and 68% technical replicate

reproducibility respectively, with higher false positive rates at a peptide level of 4.3% and

2.3% for non nitrogen fixing and nitrogen fixing cultures respectively.

Normalized spectral abundance factor (NSAF) values were calculated for each

protein detected. Proteins were grouped into functional categories and the NSAF values

totaled for each protein within a given category. Percentages of the proteome devoted to

each functional category based on the totaled NSAF values for each protein within the

category were calculated and results are given in Table 3.2 for Sp245 and Table 3.3 for

Sp7 strain. The number of proteins contributing to each category, and the percentage of

those expressed proteins based on total spectral abundance (ΣNSAF) of all proteins

within a given category is tabulated with results as shown. Visual representation of the

percentage of the expressed proteins devoted to each functional category based on the

totaled NSAF values (ΣNSAF) for each protein within a functional category is shown in

the pie charts of Figure 3.1 for Sp245 cultures and Figure 3.2 for Sp7 cultures.

90

Table 3. 2 Numbers and relative abundance percentages (by NSAF) of proteins identified in each Sp245 sample by functional category

Sp245 Sp245 control Nitrogen Fix -Mo Nitrogen Fix-V

FC Functional Category Description No. Common

Proteinsa

No.

Proteinsb

ΣNSAF

Percentagec

No.

Proteinsb

ΣNSAF

Percentagec

No.

Proteinsb

ΣNSAF

Percentagec

B Chromatin structure and dynamics 0 0 NA 1 1

C Energy production and Conversion 76 93 6.6% 115 8.8% 110 8.5%

D Cell Division and Chromosome Partitioning 11 21 0.5% 19 0.4% 16 0.4%

E Amino Acid Transport and Metabolism 99 133 6.5% 144 10.5% 137 11.0%

F Nucleotide Transport and Metabolism 31 40 2.6% 42 1.9% 42 2.0%

G Carbohydrate Transport and Metabolism 40 54 4.3% 56 4.0% 50 4.2%

H Coenzyme Metabolism 29 60 1.5% 52 1.2% 44 1.1%

I Lipid Metabolism 20 32 1.3% 33 1.8% 30 1.8%

J Translation, Ribosomal Structure and Biogenesis 111 127 36.8% 122 26.9% 121 26.0%

K Transcription 35 53 3.9% 49 2.9% 47 3.0%

L DNA Replication, Recombination and Repair 19 41 2.7% 24 2.5% 28 3.0%

M Cell Envelope Biogenesis, Outer Membrane 23 47 1.6% 41 1.5% 32 1.3%

N Cell Motility and Secretion 15 26 0.3% 44 0.6% 28 0.4%

O

Posttranslational Modification, Protein turnover,

Chaperones 49 58 6.7% 57 5.6% 60 5.2%

P Inorganic Ion Transport and Metabolism 38 47 3.2% 57 5.7% 56 6.2%

Q

Secondary metabolites biosynthesis, transport

and catabolism 20 28 1.2% 35 1.9% 28 1.9%

R General Function Prediction Only 59 95 3.1% 94 3.0% 80 3.1%

S Function Unknown 53 77 3.1% 94 4.8% 85 4.8%

T Signal Transduction Mechanisms 21 32 1.4% 51 2.2% 43 2.3%

U Intracellular trafficking and secretion 6 10 0.2% 8 0.2% 6 0.2%

V Defense mechanisms 0 4 0.05% 1 0.01% 2 0.02%

AA Hypotheticals with similarity or homology to known 45 79 5.8% 84 6.7% 81 6.3%

BB Hypothetical 74 132 6.1% 149 6.7% 142 7.2%

TOTAL 874 1289 99.5% 1372 99.8% 1268 99% a Number of common proteins identified in all growth states b Number of proteins identified belonging to the functional category in that growth state c NSAF values were totaled for proteins in each individual functional category (ΣNSAF) to determine the percentage of the total expressed proteome within each functional

category

91

Figure 3. 1 Proteome expression profile by relative abundance (NSAF) in Sp245 cells

Pie charts were constructed by first totaling the NSAF values (ΣNSAF) of proteins found within each functional category, then determining what percentage

of the total ΣNSAF of all proteins detected was devoted to each individual functional category. From this it is clear that nitrogen fixing cells devote less of

the overall expressed proteome energy to translation and ribosomal biosynthesis (J) and more energy to energy production and conversion (C) and amino

acid transport and metabolism (E).

A) Sp245 Optimal growth control B) Sp245 N2-fix molybdate C) Sp245 N2 fix vanadium

92

Table 3. 3 Numbers and percentages of relative abundance of proteins (by NSAF) identified in each Sp7 sample by functional category

Sp7 Sp7 Sp7-N2Fix

FC Functional Category Description

No. Common

Proteinsa

No.

Proteinsb

ΣNSAF Percentage

c

No.

Proteinsb

ΣNSAF Percentage

c

B Chromatin structure and dynamics 1 1 NA 1 NA

C Energy production and Conversion 80 90 6.3% 109 9.0%

D Cell Division and Chromosome Partitioning 14 18 0.4% 16 0.4%

E Amino Acid Transport and Metabolism 110 135 6.5% 137 13.9%

F Nucleotide Transport and Metabolism 33 37 2.6% 42 2.2%

G Carbohydrate Transport and Metabolism 40 49 3.9% 55 5.3%

H Coenzyme Metabolism 33 45 1.2% 48 1.5%

I Lipid Metabolism 26 35 1.2% 34 1.8%

J Translation, Ribosomal Structure and Biogenesis 113 126 41.1% 119 25.9%

K Transcription 35 45 4.2% 37 2.8%

L DNA Replication, Recombination and Repair 25 33 3.0% 34 2.4%

M Cell Envelope Biogenesis, Outer Membrane 22 37 1.6% 38 2.1%

N Cell Motility and Secretion 10 17 0.2% 19 0.3%

O Posttranslational Modification, Protein turnover, Chaperones 49 59 6.6% 57 5.5%

P Inorganic Ion Transport and Metabolism 32 43 3.7% 49 6.3%

Q Secondary metabolites biosynthesis, transport and catabolism 21 24 0.6% 33 1.9%

R General Function Prediction Only 54 75 2.5% 76 3.7%

S Function Unknown 42 61 2.5% 73 2.9%

T Signal Transduction Mechanisms 21 31 1.3% 37 2.2%

U Intracellular trafficking and secretion 7 11 0.3% 7 0.3%

V Defense mechanisms 2 4 0.05% 4 0.05%

AA Hypotheticals with similarity or homology to known 41 64 6.3% 61 4.9%

BB Hypothetical 52 77 3.8% 95 4.6%

TOTAL 863 1117 99.8% 1181 99.9% a Number of proteins identified in each growth state b Number of proteins identified in each functional category in that growth state c NSAF values were totaled for proteins in each individual functional category (ΣNSAF) to determine the percentage of the total expressed proteome within each functional

category

93

Figure 3. 2 Sp7 protein expression by functional category in A) Sp7 optimal growth controls and B) Sp7 nitrogen fixing cultures

Pie charts were constructed by first totaling the NSAF values of proteins found within each functional category (ΣNSAF), then determining what percentage

of the total NSAF of all proteins detected was devoted to each individual functional category. From this it is clear that nitrogen fixing cells devote less of the

overall expressed proteome energy to translation and ribosomal biosynthesis (J) and more energy to energy production and conversion (C) and amino acid

transport and metabolism (E).

A) Sp7 Optimal growth control B) Sp7 Nitrogen fixing

Sp7 N2 fixing

AA

5% BB

5%

C

9%

E

14%

F

2%

G

5%

H

1%

I

2%

J

26%

K

3%

L

2%

M

2%

O

5%

P

6%

Q

2%

R

4%S

3%

T

2%

Sp7 Non N2 Fix

AA

6% BB

4%C

6%

E

7%

F

3%

G

4%

H

1%

I

1%

J

41%

M

2%

R

3%

S

3%

K

4%

L

3%

P

4%

O

7%

94

Expression of nitrogenase structural components

Examination of the genome structure shows clusters of nitrogen fixation genes in

addition to the genes encoding the structural components of nitrogenase. The majority of

the structural genes and associated synthesis proteins are found within a 30kb region of

DNA, with other associated proteins scattered throughout the genome. All nitrogenase

structural components, NifH, NifD, and NifK were detected in all Sp245 nitrogen fixing

cultures (Table 3.4), with nitrogenase alpha chain having 40% sequence coverage. None

of these components were detected in cultures that were grown with ammonium,

consistent with the induction of these proteins only under nitrogen fixation conditions.

The Sp7 genome also shows two nitrogen fixation clusters in a 30kb region of

Sp7 DNA [98] with additional nitrogen fixation related proteins being scattered

throughout the genome [135]. Table 3.5 contains a list of those proteins involved in

nitrogen fixation that were detected in the Sp7 proteomes. Interestingly, only NifH

subunit of nitrogenase (abra_1247) was found in Sp7 nitrogen-limited cultures and other

structural components could not be detected. This is not surprising since a BLASTx

[134] search of the translated coding sequences of the Sp7 strain nif operon revealed only

a 32% and 27% sequence similarity to the Sp245 molybdate nitrogenase alpha

(abra_1246) and beta (abra_1245) chains, respectively. The Sp7 translated coding

sequence for the Fe protein, NifH, however, showed a 100% sequence similarity to the

Sp245 NifH sequence, while other translated coding sequences in the Sp7 nif operon

showed between 94% and 99% sequence similarity to those found in the Sp245 genome

sequence.

95

Table 3. 4 Nitrogenase components and related proteins necessary for nitrogen fixation detected in both nitrogen fixing and optimal growth control

Sp245 cultures

Gene Name Description Sp245-N2Fix- Mo Sp245-N2Fix-V Sp245

No.

Replicatesa

Average

NSAFb

No.

Replicatesa

Average

NSAFb

No.

Replicatesa

Average

NSAFb

abra_0572 nitrogen regulatory protein P-II 2 9.7E-03 2 8.6E-03 2 1.0E-03

abra_1231 cysteine desulfurase NifS 2 2.5E-04 2 2.4E-04 0 NDc

abra_1232 NifU Fe-S cluster assembly protein NifU 2 1.7E-03 2 1.3E-03 0 ND

abra_1236 ferredoxin III, nif-specific 2 5.9E-04 2 7.9E-04 0 ND

abra_1237 protein of unknown function DUF683 2 3.1E-03 2 2.8E-03 0 ND

abra_1238 nitrogen fixation protein 1 3.6E-04 2 1.7E-04 0 ND

abra_1239 nitrogen fixation protein NifX 2 5.8E-04 2 1.6E-03 0 ND

abra_1245 nifK nitrogenase molybdenum-iron protein beta chain 2 1.1E-03 2 1.3E-03 0 ND

abra_1246 nitrogenase molybdenum-iron protein alpha chain 2 5.1E-04 2 6.0E-04 0 ND

abra_1247 nifH nitrogenase iron protein 2 1.0E-02 2 1.1E-02 0 ND

abra_1251 nitrogenase-associated protein 1 4.6E-04 0 ND 0 ND

abra_1254 conserved hypothetical protein 2 2.1E-03 2 3.2E-04 0 ND

abra_1257 nitrogen fixation protein FixT 2 2.1E-03 2 2.2E-03 2 4.1E-04

abra_1350 NifQ family protein 1 1.5E-04 2 2.8E-04 0 ND

abra_3965 ptsN PTS IIA-like nitrogen-regulatory protein PtsN 2 1.0E-03 2 1.1E-03 2 1.4E-03

abra_0535 glnD protein-P-II uridylyltransferase 0 ND 0 ND 1 3.0E-05

abra_1564 GlnA glutamine synthetase, type I 2 7.2E-04 2 4.9E-04 2 9.4E-04

abra_3054 ntrC nitrogen regulation protein NR(I) 1 7.0E-05 0 ND 0 ND a Number indicates how many replicates contained the identified protein b Measure of relative abundance of each component based on spectral abundance c ND = not detected in these cultures

96

Table 3. 5 Nitrogenase components and related proteins necessary for nitrogen fixation detected in both nitrogen fixing and optimal growth control

Sp7 cultures

Gene Name Description Sp7- N2 fix Sp7

No.

Replicatesa

Average

NSAFb

No.

Replicatesa

Average

NSAFb

abra_0572 nitrogen regulatory protein P-II 2 4.5E-03 2 5.7E-04

abra_1247 nifH nitrogenase iron protein 1 1.9E-04 0 ND

abra_1257 nitrogen fixation protein FixT 2 8.5E-04 0 ND

abra_3965 ptsN PTS IIA-like nitrogen-regulatory protein PtsN 2 6.5E-04 2 1.2E-03

abra_0535 glnD protein-P-II uridylyltransferase 2 4.0E-05 0 ND

abra_1564 GlnA glutamine synthetase, type I 2 2.5E-03 2 8.9E-04

abra_3054 ntrC nitrogen regulation protein NR(I) 0 ND 0 a Number indicates how many replicates contained the identified protein b Measure of relative abundance of each component based on spectral abundance c ND = not detected in these cultures

97

Differentially expressed proteins

Direct comparisons of relative abundance of a given protein in the proteome were

done through a direct comparison of NSAF values of a given protein in the nitrogen

fixing proteome with the relative abundance of that same protein within the optimally

grown culture (in presence of ammonium and with oxygen and so “non nitrogen fixing”)

proteome. Up-regulated fold change of individual proteins was determined by the

formula (NSAF)N2fix/(NSAF)c where (NSAF)N2fix is the NSAF value for a given protein

under nitrogen fixing conditions, and (NSAF)c is the NSAF value for the corresponding

protein in optimal growth control conditions. Levels of down-regulation were

determined by (NSAF)c / (NSAF)N2fix. Levels of up- and down-regulation for proteins

detected in Sp245 cultures are given in Tables 3.6 and 3.7, respectively, while those for

Sp7 strain are given in Tables 3.8 and 3.9, respectively. Fold-change levels determined

as above were tabulated for a direct comparison between Sp245 and Sp7, and results

presented Tables 3.10 and 3.9 for up- and down-regulated proteins, respectively.

Discussion

Percentages of proteins detected in other studies of bacterial proteomes range

from 24-34% of gene coding sequences detected in proteomic studies [25, 26, 30, 32, 66],

slightly higher than the proteome coverage from bacterial cultures in our study. Further,

the numbers of detected proteins in R. palustris control proteomes is higher in a 5-step

MudPIT procedure than the numbers of identified proteins in these studies. This could be

due to several different reasons. First, peptides from abundant ribosomal proteins eluted

in every chromatographic step, thereby decreasing the apparent dynamic range of the

instrument due to the inability to sample peptides from other proteins that are of lower

98

Table 3. 6 Fold change of proteins found to be up-regulated in Sp245 nitrogen fixing cultures in comparison to optimal growth Sp245 controls

Gene Product FuncCat

Avg Fold

Change Upreg

Upreg

N2Fix-Mo

Upreg

N2Fix-V

abra_0588 Rubrerythrin AA 6.1 6.1 6.1

abra_3154 UspA domain protein AA 8.1 7.5 8.6

abra_2681 conserved hypothetical protein BB 15.6 12.2 19.0

abra_4017 conserved hypothetical protein BB 5.8 5.4 6.3

abra_0347 cytochrome c oxidase, cbb3-type, subunit II C 4.7 5.1 4.4

abra_1181 molybdopterin oxidoreductase C 5.7 6.3 5.1

abra_1191 H+transporting two-sector ATPase B/B' subunit C 4.3 5.8 2.9

abra_4904 Pyridoxal 4-dehydrogenase C 6.6 6.0 7.2

abra_0572 nitrogen regulatory protein P-II E 8.5 9.0 8.0

abra_1798 Extracellular ligand-binding receptor E 5.1 4.5 5.8

abra_1842 Extracellular ligand-binding receptor E 5.1 NDa 5.0

abra_1868 cationic amino acid ABC transporter, periplasmic binding protein E 8.8 8.5 9.1

abra_4656 putative branched-chain amino acid ABC transport system substrate-binding protein E 4.9 4.0 5.9

abra_2710 PQQ-dependent dehydrogenase, methanol/ethanol family G 8.0 9.1 7.0

abra_2756 elongation factor G domain IV J 10.8 12.3 9.3

abra_0430 periplasmic solute binding protein P 4.2 3.0 5.2

abra_0617 Cytochrome-c peroxidase P 8.1 6.9 9.1

abra_2041 3'(2'),5'-bisphosphate nucleotidase P 4.1 5.2 3.0

abra_2753 Ferritin Dps family protein P 4.6 5.3 3.9

abra_1257 nitrogen fixation protein FixT Q 5.3 5.2 5.5

abra_6359 acetoacetyl-CoA reductase Q 5.3 5.6 5.1

abra_1884 TRAP transporter solute receptor, TAXI family R 6.9 5.3 8.6

abra_0139 conserved hypothetical protein S 46.2 49.6 42.8

abra_0141 conserved hypothetical protein S 29.9 33.6 26.3

abra_1003 protein of unknown function DUF1476 S 7.3 4.4 10.1

abra_1380 protein of unknown function DUF1013 S 5.9 4.5 7.3 aNot detected to be upregulated in nitrogen fixing cultures when compared to non nitrogen fixing control cultures

99

Table 3. 6 (cont) Fold change of proteins found to be upregulated in Sp245 nitrogen fixing cultures in comparison to optimal growth Sp245 controls

Gene Product FuncCat

Avg Fold

Change Upreg

Upreg

N2Fix-Mo

Upreg

N2Fix-V

abra_1485 protein of unknown function DUF344 S 7.7 6.2 9.2

abra_3461 conserved hypothetical protein S 5.8 5.9 5.7

abra_4625 conserved hypothetical protein S 4.7 5.9 3.6

abra_0303 CBS domain containing protein T 12.7 11.5 13.9 aNot detected to be downregulated in nitrogen fixing cultures when compared to non nitrogen fixing control cultures

100

Table 3. 7 Fold change of proteins identified in Sp245 nitrogen fixing cultures to be down-regulated over Sp245 optimal growth control cultures

Gene Product FuncCat Avg Fold Change

Downregulated

Downreg

Sp245

N2Fix_Mo

Downreg

Sp245

N2Fix_V

abra_0403 helix-turn-helix domain protein AA 9.7 NDa 9.7

abra_2568 helix-turn-helix domain protein AA 7.3 9.6 5.1

abra_0639 hypothetical protein BB 6.9 ND 6.9

abra_1792 hypothetical protein BB 6.0 3.4 6.0

abra_3834 conserved hypothetical protein BB 12.2 ND 12.2

abra_6669 periplasmic nitrate reductase, large subunit C 11.2 11.2 ND

abra_4719 Glu/Leu/Phe/Val dehydrogenase E 4.8 5.3 4.3

abra_4885 Substrate-binding region of ABC-type glycine betaine transport system E 14.0 ND 14.0

abra_1004 adenylosuccinate lyase F 6.2 7.1 5.3

abra_1950 glyceraldehyde 3-phosphate dehydrogenase G 5.9 6.2 5.6

abra_1731 thiamine biosynthesis protein ThiC H 10.0 ND 10.0

abra_2597 Polyprenyl synthetase H 4.0 5.9 2.1

abra_3047 lipoic acid synthetase H 5.3 5.3 ND

abra_3999 adenosylhomocysteinase H 5.3 3.3 7.3

abra_0152 sun protein J 4.3 3.5 5.2

abra_0364 ribosomal protein S20 J 5.6 4.9 6.3

abra_0947 ribosomal protein L21 J 5.1 5.5 4.8

abra_4091 translation initiation factor IF-1 J 4.0 2.4 5.7

abra_4317 glutamyl-tRNA(Gln) amidotransferase, C subunit J 4.6 5.5 3.7

abra_2337 RNA polymerase sigma factor RpoD K 6.2 3.7 8.5

abra_3468 Cold-shock protein DNA-binding K 3.8 5.2 2.3

abra_0693 DEAD/DEAH box helicase domain protein L 9.8 ND 9.8

abra_3041 DEAD/DEAH box helicase domain protein L 7.5 9.8 5.1

abra_0563 chaperone protein DnaJ O 8.1 7.6 8.6

abra_2258 Redoxin domain protein O 6.2 5.7 6.6

abra_3139 FeS assembly protein SufD O 4.9 4.4 5.5 aNot detected to be downregulated in nitrogen fixing cultures when compared to non nitrogen fixing control cultures

101

Table 3.7 (cont) Fold change of proteins identified in Sp245 nitrogen fixing cultures to be down-regulated over optimal growth control cultures

Gene Product FuncCat Avg Fold Change

Downregulated

Downreg

Sp245

N2Fix_Mo

Downreg

Sp245

N2Fix_V

abra_0350 SAM-dependent methyltransferase R 5.2 3.6 6.8

abra_3520 cobalamin biosynthesis protein CobW R 5.0 5.1 5.0

abra_5399 amidohydrolase R 9.4 9.4 NDa

abra_3007 signal recognition particle-docking protein FtsY U 4.6 5.6 3.6 aNot detected to be downregulated in nitrogen fixing cultures when compared to non nitrogen fixing control cultures

102

Table 3. 8 Fold change of proteins up-regulated in Sp7 nitrogen fixing cultures in comparison to Sp7 optimal growth control cultures

Name Description UpregSp7-N2Fix a FuncCat

abra_2702 cytochrome c class I 14.8 C

abra_5376 cytochrome c class I 6.5 C

abra_0362 extracellular solute-binding protein family 1 8.9 E

abra_0572 nitrogen regulatory protein P-II 7.9 E

abra_3314 extracellular solute-binding protein family 3 9.3 E

abra_3517 Extracellular ligand-binding receptor 6.0 E

abra_3555 Extracellular ligand-binding receptor 7.1 E

abra_4115 ABC transporter related 8.9 E

abra_4656 putative branched-chain amino acid ABC transport system substrate-binding protein 6.5 E

abra_6441 4-phytase 26.2 E

abra_3820 extracellular solute-binding protein family 1 6.0 G

abra_4586 Methyltransferase type 11 13.7 H

abra_4652 acyl-CoA dehydrogenase domain protein (lipid metabolism) 12.0 I

abra_1829 Lytic transglycosylase catalytic 5.9 M

abra_0617 Cytochrome-c peroxidase 13.4 P

abra_2519 NMT1/THI5 like domain protein 5.5 P

abra_2753 Ferritin Dps family protein 9.2 P

abra_4287 sulfate ABC transporter, periplasmic sulfate-binding protein 7.7 P

abra_3090 TRAP dicarboxylate transporter- DctP subunit 9.1 Q

abra_4158 TRAP dicarboxylate transporter- DctP subunit 13.1 Q

abra_6186 TRAP dicarboxylate transporter- DctP subunit 16.4 Q

abra_6359 acetoacetyl-CoA reductase 5.6 Q

abra_4265 beta-lactamase domain protein 6.9 R

abra_0139 conserved hypothetical protein 9.2 S

abra_1485 protein of unknown function DUF344 9.9 S

abra_0303 CBS domain containing protein 14.8 T

abra_0588 Rubrerythrin 5.2 AA

abra_2770 serine/threonine protein kinase 6.5 AA

abra_3660 Tetratricopeptide TPR_2 repeat protein 5.3 AA a Numbers represent fold change of upregulation in nitrogen fixing cultures

103

Table 3. 9 Fold Change of Proteins down-regulated in Sp7 and Sp245 nitrogen fixing cultures in comparison to optimal growth control cultures

Gene Product FuncCat

Fold Change

Downreg in N2Fix

Downreg

Sp7N2Fix

Downreg

Sp245N2Fix_Mo

abra_2568 helix-turn-helix domain protein AA 9.5 Ca 9.5

abra_6669 periplasmic nitrate reductase, large subunit C 11.2 2.6 11.2

abra_2185 phosphoadenosine phosphosulfate reductase E 7.2 7.2 Ca

abra_4719 Glu/Leu/Phe/Val dehydrogenase E 5.3 Ca 5.4

abra_1004 adenylosuccinate lyase F 7.2 Ca 7.2

abra_1950 glyceraldehyde 3-phosphate dehydrogenase G 6.2 Ca 6.2

abra_2597 Polyprenyl synthetase H 5.9 Ca 5.9

abra_3047 lipoic acid synthetase H 5.3 2.6 5.3

abra_0947 ribosomal protein L21 J 5.4 5.2 5.5

abra_4317 glutamyl-tRNA(Gln) amidotransferase, C subunit J 5.5 NDb 5.5

abra_3468 Cold-shock protein DNA-binding K 5.2 Ca 5.2

abra_3041 DEAD/DEAH box helicase domain protein L 9.8 2.0 9.8

abra_0563 chaperone protein DnaJ O 7.5 4.2 7.5

abra_2258 Redoxin domain protein O 5.7 3.8 5.7

abra_4879 OsmC family protein O 7.8 Ca 7.8

abra_0593 ThiJ/PfpI domain protein R 9.5 9.5 4.8

abra_3520 cobalamin biosynthesis protein CobW R 5.1 2.3 5.1

abra_5399 amidohydrolase R 9.4 NDb 9.4

abra_3007 signal recognition particle-docking protein FtsY U 5.6 2.2 5.6 a C= Control cultures only, protein was not detected in nitrogen fixing cultures b ND = Protein was not detected in nitrogen fixing cultures

104

Table 3. 10 Fold Change of common proteins up-regulated in Sp7 and Sp245 molybdate nitrogen fixing cultures in comparison to each respective

optimal growth control culture

Name Description FuncCat

Avg Fold

Change Upreg

N2 Fix

Upreg

Sp245N2fix

Upreg

Sp7N2Fix

abra_0588 Rubrerythrin AA 5.7 6.2 5.2

abra_2770 serine/threonine protein kinase AA 6.5 2.0 6.5

abra_3154 UspA domain protein AA 7.5 7.5 4.8

abra_2681 conserved hypothetical protein BB 12.2 12.1 NFa

abra_4017 conserved hypothetical protein BB 5.4 5.4 NDb

abra_0347 cytochrome c oxidase, cbb3-type, subunit II C 5.1 5.1 NFa

abra_1181 molybdopterin oxidoreductase C 6.3 6.3 2.2

abra_1191 H+transporting two-sector ATPase B/B' subunit C 5.7 5.7 NFa

abra_2702 cytochrome c class I C 14.8 NFa 14.8

abra_4904 Pyridoxal 4-dehydrogenase C 6.0 6.0 4.8

abra_5376 cytochrome c class I C 6.5 NFa 6.5

abra_0362 extracellular solute-binding protein family 1 E 8.9 NFa 8.9

abra_0572 nitrogen regulatory protein P-II E 8.4 9.0 7.9

abra_1868 cationic amino acid ABC transporter, periplasmic binding protein E 8.5 8.5 2.5

abra_3314 extracellular solute-binding protein family 3 E 9.3 NFa 9.4

abra_3517 Extracellular ligand-binding receptor E 6.1 3.3 6.0

abra_4115 ABC transporter related E 8.9 4.3 8.9

abra_4656

putative branched-chain amino acid ABC transport system substrate-binding

protein E 6.5 4.0 6.5

abra_6441 4-phytase E 26.2 NFa 26.2

abra_3555 Extracellular ligand-binding receptor E 7.2 NFa 7.1

abra_2710 PQQ-dependent dehydrogenase, methanol/ethanol family G 9.2 9.2 NFa

abra_3820 extracellular solute-binding protein family 1 G 6.0 NDb 6.0

abra_4586 Methyltransferase type 11 H 13.7 NFa 13.7

abra_4652 acyl-CoA dehydrogenase domain protein I 12.0 NDb 12.0

a NF = Protein was detected only in nitrogen fixing cultures

b ND = Protein was not detected to be upregulated in nitrogen fixing cultures

105

Table 3.10 (cont) Fold Change of common proteins up-regulated in Sp7 and Sp245 molybdate nitrogen fixing cultures in comparison to each

respective optimal growth control culture

Name Description FuncCat

Avg Fold

Change Upreg

N2 Fix

Upreg

Sp245N2fix

Upreg

Sp7N2Fix

abra_2756 elongation factor G domain IV J 12.3 12.3 NFa

abra_1829 Lytic transglycosylase catalytic M 5.8 NFa 5.9

abra_2519 NMT1/THI5 like domain protein P 5.5 NFa 5.5

abra_2753 Ferritin Dps family protein P 7.2 5.3 9.2

abra_4287 sulfate ABC transporter, periplasmic sulfate-binding protein P 7.7 2.4 7.7

abra_1257 nitrogen fixation protein FixT Q 5.2 5.2 NFa

abra_3090 TRAP dicarboxylate transporter- DctP subunit Q 9.1 NFa 9.1

abra_4158 TRAP dicarboxylate transporter- DctP subunit Q 13.1 NFa 13.1

abra_6186 TRAP dicarboxylate transporter- DctP subunit Q 16.4 NDb 16.4

abra_6359 acetoacetyl-CoA reductase Q 5.6 5.6 5.6

abra_1884 TRAP transporter solute receptor, TAXI family R 5.3 5.3 4.3

abra_4265 beta-lactamase domain protein R 6.9 NFa 6.9

abra_0139 conserved hypothetical protein S 29.4 49.6 9.2

abra_0141 conserved hypothetical protein S 33.6 33.6 NFa

abra_1485 protein of unknown function DUF344 S 8.1 6.2 9.9

abra_3461 conserved hypothetical protein S 5.9 5.9 NFa

abra_4625 conserved hypothetical protein S 5.9 5.9 2.9

abra_0303 CBS domain containing protein T 13.2 11.5 14.8 a NF = Protein was detected only in nitrogen fixing cultures

b ND = Protein was not detected to be upregulated in nitrogen fixing cultures

106

abundance. Second, a number of phage sequences are known to be integrated into the

genome [128], and perhaps these are called as candidate genes but not expressed under

these growth conditions. Third, investigation of the genome sequence reveals a number

of redundant genes in the genome. Perhaps only one copy needs to be expressed, or one

version is expressed in one growth state while a different version is expressed in a

different growth state, and perhaps still others are expressed under stress conditions.

Further, strain Sp7 identifications may be low due to the use of a genome that is not

specific for that strain. However, BLASTx comparisons of the Sp7 strain sequenced

genes available in Genbank with the Sp245 genome indicate that the two strains possess

genes with a high degree of similarity with more than 95% of the available gene

sequences for Sp7 showing greater than 90% similarity to Sp245 genes. This is partly

expected since these are two strains of the same species. Only 1 gene (Sp7 nirS) did not

have a homolog in the Sp245 sequence. Two others (nitrogenase structural components,

alpha chain and beta chain of the molybdate nitrogenase) showed 27% and 32%

similarity to those found within the Sp245 genome. Therefore, although some proteins

may not be detected in the Sp7 strain due to sequence differences in the Sp245 genome, it

is likely to be a small fraction of these, as reflected in the slight differences in the

numbers of identifications and the percentage of the proteome detected between the two

strains. Further, the proteins that are detected and identified in Sp7 in the approach used

here are likely to be accurate since there is a high degree of sequence similarity between

the Sp245 strain and Sp7 strain sequences.

107

Sp245 proteome evaluation

Figure 3.1 illustrates basic shifts in a number of functional categories when

cultures are grown under nitrogen fixing conditions, most notably a decrease in

percentages of proteins detected from functional category J representing those proteins

involved in translation, ribosomal structure and biogenesis. Although the actual numbers

of proteins detected in functional category J under each growth condition are similar

(Table 3.2), the decrease in proteome percentage calculated by ΣNSAF values of proteins

in this category (~37% in optimal growth control cells versus ~ 26% in nitrogen fixing

cells) suggests a lower abundance of these proteins in the overall protein expression of

the cells, which in turn suggests a slower protein synthesis and/or turnover in nitrogen

fixing cells. The entire expression profile of the detected proteome as represented by

functional categories suggests a basic change in metabolism in cultures that are grown

under conditions of nitrogen fixation (Table 3.2 and Figure 3.1). As discussed in Chapter

1, nitrogen fixation requires a substantial amount of energy expenditure, with

concomitant adjustments in metabolism. An additional strategy to conserve energy

includes transport of molecules in order to glean all available nitrogen from the

surroundings as well as to provide additional metals needed for nitrogenase structural

components [141]. The necessity of transporting molecules combined with the need for

more energy in cells that have no fixed nitrogen available [141] is reflected in the

proteome expression profile in the following ways. Under nitrogen fixation conditions, a

higher percentage of proteins were detected in functional categories involved with amino

acid (category E: ~11% versus ~6% for control cells) and inorganic ion transport and

metabolism (category P: ~6% versus 3% for optimal growth control cells), and in energy

108

production and conversion (category C, ~8.5% versus ~6% for cells grown with

ammonium) compared to protein expression levels in cells grown with ammonium.

These observations taken together support the assumption that cells adjust their

physiology and metabolism when grown under different conditions.

Nitrogen fixation proteins

Nitrogen fixation is an energetically costly endeavor for the cell, requiring 16

ATP and 8 NADH for every molecule of nitrogen converted to 2 molecules of ammonia

[142]. As mentioned in Chapter 1, conversion of nitrogen to ammonia is accomplished

by the nitrogenase enzyme, a complex containing two proteins that are encoded by the

nifHDK operon [96]. Reduction of nitrogen is accomplished by dinitrogenase, which is

encoded by nifDK, and contains an iron-molybdenum (Fe-Mo) co-factor. Dinitrogenase

reductase protein is encoded by nifH, contains an Fe co-factor, and serves to donate

electrons to dinitrogenase [96]. Transcription of genes from the nifHDK operon is tightly

controlled by several layers of regulation, all dependent upon the nitrogen nutritional

status of the cell and nitrogen availability in the environment. The promoter site driving

transcription of the nitrogenase enzymes is an rpoN promoter site, and transcription

requires the presence of NtrA (RpoN) and NifA, a transcription factor responsible for nif

gene transcription [91]. NifA is constitutively expressed in A. brasilense, and levels of

its activity are regulated by posttranslational modification of the protein and are mediated

by both ammonia and oxygen levels: high levels of ammonium and oxygen act together

to repress NifA activity [143]. In growth conditions in which nitrogen availability is low

and oxygen levels are less than 2%, NifA is active, and nitrogenase structural

components are transcribed. All structural components are found to be expressed in

109

nitrogen fixing cultures (Table 3.4) while being absent in nitrogen-replete oxygen-

optimized cultures, consistent with the expression of the nif operon in environments with

high carbon to nitrogen ratios.

A number of bacterial species including Azotobacter, Anabaena [144], and

Rhodopseudomonas [105] can synthesize an alternate nitrogenase which uses vanadium

as a co-factor under conditions where molybdate concentrations are low and vanadium is

present. Examination of the genome sequence of Sp245 reveals the presence of a vnf

operon encoding the alpha, beta and delta chains of vanadium nitrogenase. However, the

nitrogen fixing cultures supplemented with vanadium did not show evidence of

expression of a vanadium nitrogenase under the growth conditions tested in this work.

Vanadium nitrogenase contains an iron (Fe) protein with 89% sequence similarity to

molybdenum Fe-protein. The iron-vanadium (FeV) protein of vanadium nitrogenase is

encoded by vnfDKG and consists of 3 different subunits, alpha, beta and delta [144].

BLASTp [52] alignment between the vanadium nitrogenase alpha subunit (abra_6100)

and the molybdate nitrogenase alpha subunit (abra_1248) shows 37% sequence identity

between the two alpha chains, with gaps and alignments as shown in Figure 3.3. The

detected peptide sequences in both molybdate and vanadium nitrogen fixing cultures for

the molybdate nitrogenase alpha chain are in red text and underlined in Figure 3.3. An

additional peptide N-terminal to the alignment shown was also detected in both vanadium

and molybdate nitrogen fixing cultures.

Although cultures supplemented with vanadium show evidence of nitrogen

fixation, only Mo-nitrogenase was detected in cultures supplemented with vanadium

110

AB1246 24 EKSRKRRAKH--LNVLEAEAKDCGVKSNIKSIPGVMTIRGCAYAGSKGVVWGPIKDMIHI 81

+K R KH L + + +D SN +IPG ++ RGCA+ G+K V+ G +KD I +

AB6100 9 DKDIPEREKHIYLKAPDEDTRDYLPLSNAATIPGTLSERGCAFCGAKLVIGGVLKDTIQM 68

AB1246 82 SHGPVGCGYYSWSGRRNYYVGDTGVDSWGTMHF------TSDFQEKDIVFGGDKKLHKVI 135

HGP+GC Y +W +R Y D G HF ++D +E IVFGG+K+L K +

AB6100 69 IHGPLGCAYDTWHTKR--YPTDNG-------HFNMKYVWSTDMKESHIVFGGEKRLEKSM 119

AB1246 136 EEINELFPLVNGISIQSECPIGLIGDDIEAVARAKSEELGKP---VVPVRCEGFRGVSQS 192

E + P V + + + CP LIGDDI+AVA+ + +P V V C GF GVSQS

AB6100 120 HEAFDAMPDVKRMIVYTTCPTALIGDDIKAVAKKVMD--ARPDVDVFTVECPGFSGVSQS 177

AB1246 193 LGHHIANDVIRDWIFEKTEPKEGFVSTPYDVTIIGDYNIGGDAWASRILLEEIGLRVIAQ 252

GHH+ N WI EK E +++PY + IGD+NI GD + + +G++VIA

AB6100 178 KGHHVLN---IGWINEKVGTLEPEITSPYTMNFIGDFNIQGDTQLLQTYWDRLGIQVIAH 234

AB1246 253 WSGDGTLAELENTPKAKVNLIHCYRSMNYIARHMEEKFGIPWMEYNFFGPSQIAESLRKI 312

++G+GT +L +A++N+++C RS YIA +++++GIP ++ + +G + +AE +RKI

AB6100 235 FTGNGTYDDLRKMHRAQLNVVNCARSSGYIANELKKQYGIPRLDIDSWGFNYMAEGIRKI 294

AB1246 313 AALFDDTIKENAEKVIAKYQPMVDAVIAKFKPRLEGKKVMIYVGGLRPRHVVDAYH-DLG 371

A F I+E E +IA+ + + +K RL+G ++ I+ GG R H + DLG

AB6100 295 CAFFG--IEEKGEALIAEEYALWKPKLDWYKERLKGTRMAIWTGGPRLWHWTKSVEDDLG 352

AB1246 372 MEIVGTGYEFAHNDDYQRTQHYVKEGTLIYDDVTAFELEKFVEVMRPDLVASGIKEKYVF 431

+E+V +F H +D+++ +EGT DD E + +++++PD++ +G + +

AB6100 353 VEVVAMSSKFGHEEDFEKVIARGREGTYYIDDGNELEFFEIIDLVKPDVIFTGPRVGELV 412

AB1246 432 QKMGLPFRQMHSWDYSGPYHGYDGFAIFARDMDLAINNPV 471

+K +P+ H++ ++GPY G++GF ARDM A+NNP+

AB6100 413 KKQHIPYVNGHAY-HNGPYMGFEGFVNLARDMYNAVNNPL 451

Figure 3. 3 BLASTp alignment of alpha subunits of molybdate nitrogenase (AB1246) and vanadium

nitrogenase (AB6100)

BLAST alignment of molybdenum nitrogenase alpha chain (abra_1246) and vanadium nitrogenase alpha

chain (abra_6100). There is 35% sequence identity between the alpha subunits of molybdate and vanadium

nitrogenases. The peptide sequences of abra_1246 identified in Sp245 cultures are in red bold underlined

letters. An additional peptide upstream of this alignment was also detected in all Sp245 nitrogen fixing

cultures.

111

trichloride. No peptides from vanadium nitrogenase (V-nitrogenase) were identified in

vanadium nitrogen-fixing cultures, suggesting that the V-nitrogenase is not expressed or

is expressed only at very low levels under the conditions of these experiments. This

could be due to several reasons. First, V-nitrogenase is often hierarchically expressed,

and will not be transcribed in cultures where as little as 2 nM molybdate is present [103].

If Sp245 V-nitrogenase is hierarchically expressed, then even small amounts of

molybdate would affect its expression. Since metals were not chelated prior to growth, it

is possible that a high enough concentration of molybdate was present in the water with

which the growth media was made to ensure repression of the V-nitrogenase. V-

nitrogenase has low efficiency in converting nitrogen, and makes small amounts of

hydrazine in addition to ammonia, but is more effective at lower temperatures [144].

Because of this lower efficiency, it is possible that once the culture became molybdate-

limited it began to express vanadium nitrogenase but the amount was below the level of

detection in these experiments. Alternatively, the V-nitrogenase could be expressed at a

low level simultaneously with the Mo-nitrogenase, but not detected due to very low

abundance.

Growth and energy production and conversion

Comparison of the most abundant proteins and the degree of up-regulation in

nitrogen fixing cultures with those of cultures grown with ammonium suggests a shift in

basic metabolism in nitrogen fixing cells. Investigation of the most abundant proteins in

each growth condition listed in order by NSAF values (found in attached spreadsheets

Sp245N2FixData and Sp7N2FixData) suggests rapid growth in cultures grown with

ammonium. In these cultures, the most abundant proteins are heavily populated with a

112

number of different ribosomal proteins, translation elongation factors and chaperonins.

Additionally, DNA and RNA binding proteins are prevalent. Together, these data suggest

that transcription of DNA and translation of mRNA into proteins is a high priority under

these conditions and also that cells are growing rapidly, synthesizing biomass and/or with

an active protein turnover. By comparison, cells grown under nitrogen fixing conditions

include a wider variety of proteins and a more diverse representation of functional

categories among their most abundant proteins. Although ribosomal proteins, translation

elongation factors, chaperonin proteins, and DNA and RNA binding proteins are found

among the most abundant proteins in nitrogen fixing cultures, proteins in these functional

categories are not the clear majority. Instead, down-regulation of greater than 5-fold

(Table 3.7) is noted for DNA binding proteins (abra_0403, 2568, 0693, 3041, 3468),

ribosomal proteins (abra_0364, 0947), chaperone protein DnaJ (abra_0563), and RNA

polymerase sigma factor RpoD (abra_2337). Cell division proteins, including MinC,

MinD, FtsA, FtsK and FtsZ, are all down-regulated by 2-fold or greater (found in

attached spreadsheet 2fold_DownN2Fix), suggesting slower growth and cell division.

Further, a number of ribosomal proteins, chaperones and tRNA synthetases are down-

regulated by 2-fold or greater (found in attached spreadsheet 2fold_DownN2Fix),

indicating a slower protein synthesis in nitrogen fixing cells. These data suggest that

while nitrogen fixing cells are actively growing, their growth is slower, biomass synthesis

is reduced and protein turnover less active, consistent with the observation that cells grow

slower and to lower cell density under conditions of nitrogen fixation [140].

As discussed in Chapter 1, nitrogen fixation is an energetically expensive process,

and nitrogen fixing cells must address this need for more energy [141]. A number of

113

cytochrome c proteins are also found among the most abundant proteins in cells grown

under nitrogen fixation conditions, consistent with the need for more energy in order to

fix nitrogen. Three cytochrome c proteins (abra_5376, 4904, and 3146) are found to be

up-regulated by greater than 2-fold (found in attached table 2fold_UpN2Fix), while

cytochrome c oxidase (abra_0347) is up regulated by greater than 5-fold (Table 3.6) with

a concomitant increase in sequence coverage to 50%. Additional energy requirements

are met through up-regulation of ATP synthase (abra_1191, 5-fold and abra_0929, 2-

fold) and PQQ-dependent dehydrogenase methanol/ethanol family (abra_2710). Pyrrolo-

quinoline-quinone (PQQ)-dependent dehydrogenase is a periplasmic quinoprotein that

oxidizes methanol or ethanol to formaldehyde, passing the electrons derived from this

oxidation to soluble cytochrome cL. Additional reducing equivalents are provided by 2-

fold up-regulation (found in attached table 2fold_DownN2Fix) of citric acid cycle

enzymes, malate dehydrogenase (abra_2984 and 0339), and succinate dehydrogenase

(abra_2959).

Sensing energy and nitrogen status of the cell becomes very important when cells

are grown under nitrogen fixing conditions [141]. Nitrogen regulatory P-II proteins

serve as a sensor for nitrogen status in the cell, and are involved in regulatory control of

nitrogen metabolism on several different levels [91, 96, 97]. Abra_0572, annotated as

nitrogen regulatory protein P-II, was identified with over 80% sequence coverage and an

up-regulation of 8.5-fold in all nitrogen fixing cultures (Table 3.6). Additionally, all

nitrogen fixing cultures show substantial up-regulation of abra_0303, a cystathionine-

beta-synthase (CBS) domain containing protein, identified with over 80% sequence

coverage in all nitrogen fixing cultures. CBS domains are small intracellular modules

114

that pair together to form stable globular domains and are often combined with a variety

of other protein domains. They are known to bind ATP, AMP, and S-

adenosylmethionine (SAM), and as such can function as energy sensors or intracellular

sensors of the metabolic status of the cell [145], although they can also function as

sensors for osmolarity as well [146]. A BLASTp [134] search against the NCBI non-

redundant protein database revealed a 38% similarity to signal transduction protein from

Rhodospirillum rubrum ATCC 11170 (YP 428825.1). Since abra_0303 is up-regulated

by greater than 11.5-fold in all nitrogen fixing cultures, it is likely to be important in

sensing the status of energy or intracellular metabolites within the cell.

Several proteins involved in metal homeostasis or nucleotide metabolism are

found to be significantly up-regulated in all Sp245 nitrogen fixing cells (Table 3.6). For

instance, all Sp245 cultures grown under nitrogen-fixation conditions show up-regulation

of abra_0588 (rubrerythrin), abra_2753 (ferritin-Dps family protein) and abra_1485

(protein of unknown function DUF344). Rubrerythrin, up-regulated by 6-fold under

nitrogen fixing conditions, is a protein with a ferritin-like fold proposed to play a role in

metal ion binding. It belongs to the same family as Ferritin-Dps (DNA protection during

starvation protein), which is also up-regulated in Sp245 nitrogen fixing cultures, and is

also involved in non-heam storage of iron. Abra_1485 contains a domain structure

similar to nucleoside triphosphate hydrolases containing a P-loop, and members of this

family are known to be involved in purine and pyrimidine metabolism [147]. It further

shows sequence similarities of greater than 65% to a number of hypothetical proteins and

polyphosphate kinases from other bacterial species.

115

Molecular transport and metabolism shifts under nitrogen fixing conditions

Also among the most abundant proteins in cells grown under nitrogen fixing

conditions are a number of extracellular ligand binding domain proteins and molecular

transport proteins. These proteins are likely involved in more efficient use of available

nutrients such as amino acids or metals, consistent with expected changes in metabolism

between these 2 conditions [148, 149]. Of particular interest in molecular transport are

two conserved hypothetical proteins, abra_0139 and abra_0141, shown to be up-regulated

by greater than 30-fold under conditions of nitrogen fixation in Sp245 strain (Table 3.6),

suggesting they play major roles in the physiology of cells under these conditions.

Sequence coverage of abra_0139 (379 amino acid length) increases from 23% in

ammonium-replete cultures to 67% in nitrogen fixing cultures, while sequence coverage

increases from 28% to 78% for abra_0141 (323 amino acids). The ORFS encoding

abra_0139 and abra_0141 are not genetically linked and are not contained with an

operon. They are located 1944 nucleotides from one another in the genome sequence,

and are transcribed in opposite directions (Figure 3.4). Abra_0139 is flanked by a

methyl-accepting chemotaxis protein (abra_0140) with a start site downstream 157

nucleotides and an operon containing dimethylmenaquinone methyltransferase

(abra_0137) and an auxin efflux carrier (abra_0138) which ends 391 nucleotides

upstream, while abra_0141 is preceded by a response regulator receiver modulated metal-

dependent phosphohydrolase which ends 248 nucleotides upstream of its start site. The

close proximity of an MCP to these hypothetical proteins suggests that both proteins may

function in response to the same signaling pathway.

116

Figure 3. 4 Schematic representation of genomic region surrounding Abra_0139 and Abra_0141

Abra_0139 and abra_0141 are both hypothetical proteins containing BUG (Bordatella Uptake Gene) domains, thought to be important in solute binding and

uptake. Sp245 nitrogen fixing cultures show a large degree of upregulation of these proteins. The close proximity of auxin efflux carrier to abra_0139 is

intriguiging because of the known secretion of plant hormones in Azospirillum species associated with plants.

methyltransferase auxin efflux carrier abra_139 MCP

phosphohydrolaseabra_141

90 ntds 391 ntds 157 ntds

248 ntds

117

Analysis of abra_0139 and abra_0141 by BLASTp against the NCBI non-

redundant protein database indicate that both proteins have a 51% and 54% sequence

similarity to a putative secreted protein from Bordetella petrii and 49% and 50%

sequence similarity to a probable extra-cytoplasmic solute receptor from Ralstonia

eutropha H16, respectively. Aligning the amino-acid sequences of abra_0139 and

abra_0141 using BLASTp [52] (Figure 3.5) indicates that they share 66% overall

sequence similarity. Interestingly, abra_0141 possesses a periplasmic binding domain at

its N-terminal end that is not found in abra_0139 while both proteins possesses a C-

terminal BUG (Bordetella uptake gene) domain. BUG domains are widespread in

bacteria, and seem especially abundant in the genomes of soil bacteria. Although their

function has not been characterized, BUG domain proteins are suggested to be involved

in nutrient transport [150]. Average up-regulation of 46-fold for abra_0139 and 30-fold

for abra_0141 in Sp245 nitrogen fixing cultures suggests that both may be important for

nutrient uptake or transport in nitrogen fixing cells.

Interestingly, periplasmic domains of ABC transporter systems and extracellular

ligand-binding receptors are also found to be abundant in nitrogen fixing cultures. An

extracellular ligand-binding receptor, abra_1798, is the most abundant protein found in

all cultures grown under nitrogen fixation conditions, and is up-regulated by 4.5- and 5.8-

fold in Sp245 molybdate and vanadium nitrogen fixing cultures, respectively (Table 3.6).

This protein is not among the most abundant proteins in cultures grown with ammonium,

suggesting an important function under nitrogen fixation conditions. A BLASTp [52,

134] search against the NCBI non-redundant database (http://blast.ncbi.nlm.nih.gov/)

using the translated coding sequence revealed that abra_1798 shows 75% similarity to the

118

Figure 3. 5 BLASTp alignment of protein sequences for hypothetical proteins abra_0139 and

abra_0141

Hypothetical proteins Abra_0139 and Abra_0141 both contain a BUG (Bordatella Uptake Gene) domain

and are shown to be highly upregulated in nitrogen fixing conditions, suggesting they are important for

nutrient uptake under nitrogen fixing conditions. BLASTp analysis of the two sequences shows 67%

identity between the two with 81% positives.

119

periplasmic binding domain of ABC-type branched-chain amino acid transport systems

of Magnetospirillum magnetotacticum, a phylogenetically close relative of A. brasilense.

Searches against the NCBI conserved domain database [151] revealed that abra_1798

possesses conserved sequences matching to protein superfamily of periplasmic ligand-

binding domains of ABC-type transporters with a domain structure of the LivK family.

LivK is a periplasmic leucine-, isoleucine-, and valine-specific-binding protein involved

in leucine transport, consistent with the suggestion that under conditions of nitrogen

fixation, cells expend more of their energy in transporting molecules (Table 3.2).

An additional extracellular ligand-binding receptor (abra_5402) is also found

within the most abundant proteins in all nitrogen fixing cultures. Analysis of closest

homologs of abra_5402 in the NCBI microbial database using BLASTp [52] indicates

that abra_5402 has high sequence similarity to the extracellular ligand-binding domain of

ABC-type transporters, with the highest sequence similarity (47%) to an extracellular

ligand-binding receptor from Rhodospirillum rubrum that has an unknown function.

Further evidence of the need for increased molecular transport is seen in the abundance

and up-regulation of cationic amino acid ABC-transporter (abra_1868) and putative

branched-chain amino acid ABC transport system substrate binding protein (abra_4656).

Abra_1868 is found among the most abundant proteins in cells grown under nitrogen

fixing conditions and is up-regulated by 8.5-fold and 9.1-fold in molybdate and vanadium

nitrogen fixing cultures, respectively (Table 3.6). Abra_4656 is up-regulated by 5-fold,

and a number of additional ABC transporter binding proteins are up-regulated by greater

than 2-fold (found in attached table 2fold_UpN2Fix). The abundance and variety of

substrate binding proteins in nitrogen fixing cultures serves to emphasize the need for

120

nitrogen fixing cultures to import nutrients rather than expend the energy to synthesize

them. Taken together, these results suggest the need for increased molecular transport in

nitrogen fixing growth state.

Sp245 nitrogen fixing cultures show an abundance of cupin 2 barrel domain

protein (abra_3559) with a 2-fold increase in expression levels in nitrogen fixing Sp245

cultures (attached table 2fold_UpN2Fix). Cupin proteins are related to sugar storage

proteins in plant seeds, and have been shown to have a number of different roles in

bacteria including cell wall biosynthesis, secondary metabolite synthesis, stress response

and sugar metabolism [152]. Interestingly, cupin domains have been found to be present

in auxin binding proteins, where one of the five exons of auxin binding protein encodes a

region that is thought to act as a receptor for IAA [152]. Of further interest, bacterial

cupin domain proteins have been linked to oxalate metabolism. Oxalate, C2O42-

, has been

linked to either chelation and precipitation or to solubilization of metals in soils such that

the metal-oxalate compound is made available [152]. Cupin domain proteins are not

abundant in bacterial genomes [152], so the presence of this protein among the most

abundantly expressed proteins in Sp245 nitrogen fixing cultures suggests that this protein

plays an important role in this growth condition. Sp245 is known to produce more IAA

under nitrogen fixing conditions [153], and A. brasilense is known to change its cell

morphology under nitrogen fixing conditions [154], so the role of this particular protein

could be varied, involved in transport of IAA, or cell wall remodeling or chelation of

metals. This protein is not detected in Sp7 cultures at all, either because it does not exist

within the Sp7 genome, or because it is either not expressed in Sp7 under these

conditions or is expressed at a level below our detection limit. Further, the abundance of

121

this protein in Sp245 cultures and non-detection of it in Sp7 cultures suggests that these

two strains may employ different strategies for adaptation to nitrogen fixation.

Interestingly, both Sp245 molybdate and vanadium nitrogen fixing conditions

show a 7-fold up-regulation of abra_3154, a UspA domain protein (Table 3.6). Sp7

nitrogen fixing cultures also show an up-regulation of abra_3154 of 4.7-fold, suggesting

this protein is likely to be important under nitrogen fixing conditions. In E. coli,

Universal Stress Protein (Usp) A is a small cytoplasmic bacterial protein which lends

stress endurance to the cell, enhancing cell survival rate [155]. Increased expression is

noted in cells that are under stress, including starvation for carbon, nitrogen or amino

acids, where Usp family proteins protect against oxidative stress and mediate iron

homeostasis and motility/adhesion in stress conditions [156]. Activity levels of Usp

family proteins in E. coli have been shown to be dependent upon serine-threonine

autophosphorylation [155]. In A. brasilense, a mutant lacking an Usp-family protein has

been described. This Sp7 Tn5 mutant shows increased susceptibility to carbon

deprivation, decreased EPS production and impaired flocculation, but does not exhibit

sensitivity to oxidative stress agents [157]. Since flocculation has been proposed as a

strategy used by the microaerophilic Azospirillum species to control oxygen diffusion

through the cells [158], perhaps the up-regulation of UspA in nitrogen-fixing cultures

plays a role in both EPS metabolism for nitrogen fixation energy and flocculation to

protect nitrogenase integrity.

122

Differences between Sp245 molybdate versus vanadium nitrogen fixing cultures

Vanadium is a transition metal ubiquitous in soil and water, with particularly high

concentrations in the oceans. It exists in a variety of oxidation states, from V(V) to V(II),

with V(V) and V(IV) being the most commonly found in nature, and V(V), V(IV) and

V(III) all shown to exist within different organisms. [159, 160] The anionic species,

vanadate, or V(V), has the chemical formula of H2VO4- . It is soluble and is the prevalent

oxidation state found within the soil, most often found in oligomeric forms. The cationic

species, vanadyl, or V(IV), has a chemical formula VO2+. It is insoluble in water

although it is the prevalent form found in suspension in rivers. It is often complexed to

other compounds, such as the porphyrins in petroleum, so that refinement of petroleum

results in creation of vanadium sulfide [159]. The lowest oxidation state that can stay in

solution is V(III), or V3+

, but must be coordinated to another ligand (such as iron

transport protein, transferrin) or in a very acidic environment to do so. V(V) and V(IV)

function as a redox couple with an E = 1.31V [160].

The biological effect of vanadium is dependent upon its oxidation state.

Vanadate, V(V), can act as a phospho-mimetic, competing with phosphate HPO42-

while

vanadyl, V(IV) can act as a transition metal ion to compete with other divalent metal ions

[160]. Vanadate has been shown to inhibit the action of phosphatases, ribonucleases, and

ATPases by serving as a transition state analog or by replacing the terminal phosphate in

ATP, ADP or AMP and thus preventing further action by the enzymes [159, 160]. In

vitro studies indicate that it can also inhibit the action of serine and cysteine proteases

when in a sulfate salt form [161]. Vanadium has also been shown to simply act as a

metal catalyst to catalyze the oxidation of sulfates [159]. In vitro studies using

123

Azotobacter Mo-nitrogenase suggest that vanadium can be used in the more traditionally

expressed molybdate nitrogenase. In these studies, the reduced Fe protein was shown to

reduce vanadate to vanadyl, and then use this form as a substitute for magnesium in the

Mg-ATP bound to the MoFe protein [162]. Growth experiments with a number of

species of bacteria indicate that though most bacteria may not be able to reduce

vanadium, they can still tolerate more than trace amounts of vanadium, with most species

and strains tested growing well in cultures with up to 3 g/L concentration of vanadate

[163]. High concentrations, however, are toxic to some bacteria, such as Azotobacter

vinelandii, which shows decreased growth in media with concentrations greater than1µM

vanadium. [164] The effect of vanadium on Azospirillum species has not been studied.

Addition of vanadium to nitrogen fixing cultures did not appear to affect growth

of the cultures. A similar amount of growth and clumping behavior was observed in both

vanadium and molybdate cultures, with vanadium nitrogen fixing cultures growing to the

same optical density as the molybdate nitrogen fixing cultures after 48 hours of growth.

Microscopic investigation of the vanadium nitrogen fixing cultures showed actively

motile cells, with no significant difference in size or shape of individual cells. In

comparing patterns of protein expression of molybdate and vanadium nitrogen fixing

cultures when compared to a control of optimal growth control cultures, expression

profiles are similar across all functional categories (Figure 3.1 and Table 3.2). The list

of proteins detected to be up-regulated in both nitrogen fixing cultures is the same (Table

3.6) with similar levels of up-regulation for each protein. This list includes a number of

extracellular ligand binding proteins, hypothetical proteins and transporter proteins as

well as cytochrome c oxidase and ATPase subunits, all of which were discussed earlier.

124

Similar levels of up-regulation of each protein within this list suggests similar needs for

molecular transport and energy generation under both nitrogen fixation conditions.

One notable exception is extracellular ligand binding receptor abra_1842, which

is detected to be up-regulated by 5-fold in vanadium nitrogen fixing cultures but not up-

regulated at all in molybdate nitrogen fixing cultures (Table 3.6). Abra_1842 encodes a

protein with 61% similarity to a LivK family protein from Methylobacterium

radiotolerans JCM 2831, suggesting it is involved in branched chain amino acid

transport. It is part of a putative operon that encodes components of an ATP-dependent

branched chain amino acid transport system but other components of this system are not

detected to be up regulated (Figure 3.6). Upstream of this operon by 332 nucleotides is

Abra_1839, which encodes a MarR family protein. Proteins within the MarR family of

transcriptional regulators share a conserved winged helix structure, and binding to DNA

is mediated by the presence of anionic lipophilic ligands [165]. MarR dimers bind to

specific palindromic DNA sequences in order to affect transcription from Mar (multiple

antibiotic resistance) operons or oxidative stress operons or from operons involved in

aromatic catabolism [165]. They function as environmental sensors, and some have been

shown to repress transcription of those operons of which they are a part, or of operons

that are close to them in the genome structure, while others have been shown to enhance

transcription [165]. Interestingly, in vanadium nitrogen fixing cultures abra_1839 is not

detected to be down-regulated, while it is down-regulated by 4-fold in molybdate

nitrogen fixing cultures (attached spreadsheet 2Fold_DownN2fix). The proximity of this

MarR homolog, abra_1839, to transport protein abra_1842, taken together with the 5-fold

up-regulation of abra_1842 in vanadium cultures, while no up-regulation is found in

125

Figure 3. 6 Genomic region encompassing Abra_1839 MarR family transcriptional regulator Abra_1839 shows similarity to MarR family transcriptional regulator. This protein is downregulated in molybdate nitrogen fixing cultures but not in

vanadium nitrogen fixing cultures. Interestingly, this protein is in close proximity to a family of ABC-transporters involved in branched chain amino acid

transport. MarR proteins have been shown to respond to environmental changes, and to affect transcription of genes in close proximity to them. Transporter

Abra_1842 is upregulated in vanadium nitrogen fixing cultures but not in molybdate nitrogen fixing cultures, suggesting the possibility that abra_1839 may

play a role in regulation of this operon of ABC transporters in response to environmental stimuli related to the presence of vanadium in the culture..

Enoyl-CoA hydrolase Abra_1839 ABC-transporter ABC-transporter ABC-transporter IM translocator

Branched-chain amino acid transport system

abra_1842abra_1841abra_1840 abra_1843

MarR family transcriptional regulator

126

molybdate cultures where abra_1839 is down-regulated by 4-fold suggests the intriguing

possibility that abra_1839 MarR protein is acting to affect transcription of transport

protein abra_1842.

Twenty one common proteins are down-regulated in both molybdate and

vanadium Sp245 nitrogen-fixing cultures (Table 3.7), although the levels of down-

regulation are variable between cultures. Ribosomal proteins, chaperone proteins, DNA

binding proteins, and synthesis proteins show similar levels of down-regulation in both

cultures (Table 3.7), supporting the assumption that both molybdate and vanadium

nitrogen fixing cultures grow slower with concomitant decreases in DNA and protein

synthesis. Notable exceptions include helix-turn-helix binding domain protein encoded

by abra_0403, and DEAD/DEAH box helicase domain protein encoded by abra_0693,

both of which are down-regulated by more than 9-fold in vanadium cultures but are not

detected in molybdate cultures. Additional hypothetical proteins abra_0639 and 3834 are

also substantially down-regulated in vanadium cultures but not detected in molybdate

cultures. The lack of detection of these proteins in molybdate cultures indicates they

were either not expressed in molybdate nitrogen fixing cultures, or that they were

expressed at very low levels such that they were below our limit of detection, which

would imply a larger level of down-regulation of these proteins than in vanadium

cultures.

Interestingly, molybdate nitrogen fixing cultures show substantial down-

regulation of periplasmic nitrate reductase (abra_6669), but this protein is not detected in

vanadium nitrogen fixing cultures (Table 3.7). Periplasmic nitrate reductase is a

127

molybdenum-containing enzyme involved in the dissimilatory process of the nitrogen

cycle by catalyzing the reduction of nitrate (NO3-

) to either nitrite (NO2-

) or ammonia

[166]. Due to the lack of nitrate in the culture combined with the limited amount of

molybdenum and the necessity of using molybdenum for Mo-nitrogenase, down-

regulation of this protein is consistent with nitrogen fixation needs of the cell. The lack

of detection of periplasmic nitrate reductase in vanadium nitrogen fixing cultures

suggests that in a situation in which molybdate is severely limited either abra_6669 is not

expressed at all or is expressed at levels below those we can detect. This data suggests

that nitrogen cycling may be affected by the addition of vanadium to nitrogen fixing

cultures, perhaps due to limited amounts of molybdate (which may in turn lead to lower

expression levels of Mo-nitrogenase) or alternatively to less abundant or less efficient

nitrogenase activity in the vanadium cultures, or due to interference of the vanadium with

iron uptake pathways.

In addition, lipoic acid synthetase (abra_3047), which catalyzes the synthesis of

lipoic acid, a coenzyme bound to pyruvate dehydrogenase and alpha-ketoglutarate

dehydrogenase, is down-regulated by greater than 5-fold in molybdate nitrogen fixing

cultures only (Table 3.7). In Rhizobium etli, lipoic acid synthetase has been shown to be

down-regulated as it enters stationary phase when the cell energy needs are decreased

[167]. Recent studies with a mutant of B. subtilis lacking a gene for lipoic acid

synthetase have indicated that it also plays a role in branched chain fatty acid synthesis,

and can affect the composition of the outer membrane lipopolysaccharides (LPS) [168].

The above data suggests that the 5-fold down-regulation of lipoic acid in Sp245 nitrogen

fixing cells would result in a decreased output from the TCA cycle due to the reduction in

128

alpha-ketoglutarate dehydrogenase and pyruvate dehydrogenase activity. Further, down-

regulation of lipoic acid synthetase in nitrogen fixing cells may also contribute to outer

membrane remodeling such that the LPS of the outer membrane contain a greater

percentage of straight chain saturated fatty acids.

Sp7 Proteome evaluation

The proteome expression profiles of Sp7 nitrogen fixing and optimal growth

control (nitrogen replete) cells are shown in Figure 3.2. As seen in Sp245 nitrogen fixing

cultures, Sp7 also shows a shift in metabolism of nitrogen fixing cells. The most notable

differences in protein abundances by functional category are a decrease in percentages of

proteins detected from functional category J representing those proteins involved in

translation, ribosomal structure and biogenesis, and increases in functional categories C

and E representing energy production/conversion and amino acid transport/metabolism,

respectively. The number of proteins detected in categories J (translation, ribosomal

structure and biogenesis) and K (transcription) are lower in nitrogen fixing cells, while

the numbers of proteins detected in categories C (energy production and conversion) and

Q (secondary metabolite biosynthesis, transport and catabolism) are substantially higher

(Table 3.3). The protein expression profile for Sp7 nitrogen fixing cells is indicative of

slower growth and increased need for energy and molecular transport, just as in Sp245

nitrogen fixing cells. The decrease in protein abundances of those proteins involved in

translation (category J, decrease from 41% to 25.5%) support the assumption that cells

are making less protein when growing under nitrogen fixing conditions. Further, the

most abundant proteins when listed in order by NSAF values (attached table

Sp7N2Fix_Data) have far less ribosomal proteins under nitrogen fixing conditions, and

129

contain a number of extracellular ligand-binding proteins and solute binding proteins.

The increased need for molecular transport and energy in nitrogen fixing cells is also

illustrated in Sp7 nitrogen fixing cells by the increase in percentages of protein

abundance by ΣNSAF in those proteins contributing to energy production and conversion

(category C, 9% versus 6% in nitrogen replete cells) and to amino acid transport and

metabolism (category E, ~14% versus 6.5% in non nitrogen fixing cells).

Surprisingly, only one structural component of nitrogenase, abra_1247 nifH

nitrogenase Fe-protein was detected in Sp7 nitrogen fixing cultures (Table 3.5). The nif

operon from Sp7 strain was sequenced earlier [98, 169] and sequences entered into the

NCBI database (http://www.ncbi.nlm.nih.gov/). BLASTx [134] comparison of the

translated coding sequences of the Sp7 strain nif operon against the Sp245 genome

structure showed remarkable levels of similarity (94 – 100%) for the majority of Sp7

translated coding sequences with those of the coding sequences of the nif operon in

Sp245 genome. The notable exceptions to this included nitrogenase molybdate

nitrogenase alpha and beta chains (abra_1246 and abra_1245, respectively). The Sp7

translated coding sequences showed only 32% and 27% similarity, respectively, to those

same sequences from Sp245 genome. With the low degree of similarity in the alpha and

beta nitrogenase subunits, it is therefore not surprising that no peptides were detected

from these proteins in the Sp7 nitrogen fixing cultures. The Sp7 translated coding

sequence for the NifH protein, however, showed 100% sequence similarity to that from

the Sp245 genome, thus explaining the detection of the NifH protein in Sp7 cultures.

130

Growth and Energy Production

Nitrogen fixing Sp7 cells seem to have a somewhat different strategy for coping

with nitrogen fixing conditions than do Sp245 nitrogen fixing cells. Although a large

difference in percentage of proteins based on abundance by NSAF values detected that

are involved in translation and transcription are noted, only one ribosomal protein L21

(abra_0947), is found to be down-regulated by more than 5-fold (Table 3.9). However, a

number of other ribosomal proteins as well as RNA polymerase, sigma factor RpoD, and

translation initiation and termination factors are down-regulated by 2 to 3-fold (attached

spreadsheet 2fold_DownN2Fix), thus suggesting an overall decrease in protein

production or turnover. One protein, abra_0593, containing a ThiJ/PfpI domain, is found

to be down-regulated by 9.5-fold under nitrogen fixing conditions (Table 3.9). The

ThiJ/PfpI domain is found in a diverse group of proteins involved in regulation of RNA-

protein interaction, thiamine biosynthesis, or protease activity, and has also been found in

transcriptional regulators. While abra_0593 is shown to be down-regulated by 4-fold in

Sp245 nitrogen fixing cultures (attached spread sheet 2fold_DownN2fix), down-

regulation of DNA binding proteins, chaperonins and cell division proteins that is

prevalent in Sp245 nitrogen fixing cultures is not noticed in Sp7 nitrogen fixing cultures,

perhaps suggesting either a different strategy for reducing growth rate or larger levels of

down-regulation of these proteins in Sp7 cultures such that they are not detected at all

under nitrogen fixing conditions.

Increased energy requirements are met in Sp7 nitrogen fixing cells through up-

regulation of a number of respiratory chain components. ATP synthase subunits

(abra_0925 and 0928) are among the most abundant proteins in Sp7 nitrogen fixing

131

cultures. Additional ATP synthase subunits (abra_0927 and 0929) are shown to be up-

regulated by greater than 2-fold (attached spreadsheet 2fold_UpN2Fix), suggesting Sp7

cells are using this ATP synthase (among the multiple ATP synthase subunits present in

the Sp245 genome) to synthesize the large amount of ATP required for nitrogen fixation.

The up-regulation of nearly 15-fold of cytochrome c (abra_2702) and 6.5-fold of

cytochrome c (abra_5376), shown in Table 3.8, along with greater than 2-fold up-

regulation of 2 electron-transferring-flavoprotein dehydrogenases (abra_3654 and 4624)

and ubiquinol-cytochrome c reductase (abra_4336) suggests the importance of the

electron transport chain in energy production in nitrogen fixing Sp7 cells. As in Sp245

nitrogen fixing cells, reducing equivalents are provided by 2-fold up-regulation of malate

dehydrogenase (abra_2984) although no other citric acid cycle enzymes are shown to be

up-regulated (attached spreadsheet 2fold_UpN2Fix).

Molecular transport and metabolism changes under nitrogen fixation

The strategy adopted by Sp7 nitrogen fixing cells for molecular transport includes

a number of extracellular ligand-binding receptor proteins, extracellular solute binding

proteins, ABC-transporters and DctP subunits of TRAP dicarboxylate transporters, all

found within the most abundant proteins in Sp7 nitrogen fixing cultures (attached

spreadsheet Sp7_N2FixData). Extracellular solute binding protein family 1 (abra_0362

and 3820) and family 3 (abra_3314) proteins are up-regulated by 8.9-, 9.3- and 6-fold,

respectively, while extracellular ligand-binding receptor proteins (abra_3517 and 3555)

are up-regulated by 6- and 7-fold, respectively (Table 3.8). An additional 7 extracellular

solute binding proteins and 7 ligand-binding receptor proteins are up-regulated by more

than 2-fold. Both abra_0362 and abra_3820 have LysR substrate binding domains, which

132

are present in proteins involved in regulation of metabolism of carbohydrates and amino

acids, often binding co-factors or co-inducers [170]. Abra_3314 has a domain structure

similar to those found in penicillin binding proteins. Extracellular ligand-binding

receptor, abra_1798, is among the most abundant protein found in Sp7 cultures grown

under nitrogen-fixation conditions (attached spreadsheet Sp7_N2FixData), and is up-

regulated by 4-fold (attached spreadsheet 2fold_UpN2fix). As discussed earlier,

abra_1798 shows similarity to ABC-type transporters with a domain structure of the

LivK family. BUG domain protein, abra_0139, (up-regulated by greater than 40-fold in

Sp245 nitrogen fixing cultures), is up-regulated by only 9-fold in Sp7 cultures, while

abra_0141 (up-regulated by greater than 25-fold in Sp245 nitrogen fixing cultures) is

detected only in Sp7 nitrogen fixing cultures, but not in Sp7 non-nitrogen fixing cultures

so a level of up-regulation could not be determined. Taken together, the above data

suggests that Sp7 cells use a different, more diverse strategy for molecular transport than

does Sp245, perhaps reflecting different physiological abilities to cope with the metabolic

demands of nitrogen fixation.

Nitrogen fixation conditions (high C:N ratio) are known to induce

exopolysaccharides production in Azospirillum species and various mutants affected in

the composition of the exopolysaccharides have been characterized [171, 172]. Earlier

studies in strain Sp7 cells have shown that the composition of exopolysaccharide (EPS)

changes dramatically based on the growth phase and energy status of the cell [173]. In

exponential phase, the cells produce glucose-rich EPS which is later degraded and used

as a carbon source. In stationary phase, arabinose-rich EPS is produced, facilitating cell

aggregation and flocculation, a behavior thought to promote resistance from hostile

133

environmental conditions [173]. Precursor components of EPS are synthesized in the

cytoplasm, and then must be transported out to the outer membrane. This transport is

accomplished through ATP family transporters [174], an observation also consistent with

the up-regulation of a number of ATP transporters detected in Sp7 grown under nitrogen

fixation (Table 3.8).

Several proteins suggested to be localized to the inner membrane of cells were

found to be enriched in the proteomes of cells grown under conditions of nitrogen

fixation, suggesting that growth under these conditions is accompanied by major changes

in membrane-associated proteins. In Sp7 nitrogen fixing cultures, basic membrane

lipoprotein (abra_1538) and inner membrane lipoprotein A (NLPA lipoprotein,

abra_4083) are among the most abundant proteins (Attached spreadsheet

Sp7_N2FixData), showing a 2.5 and 1.5-fold increase respectively between nitrogen

fixing and non-nitrogen fixing Sp7 cultures (attached spreadsheet 2Fold_UpN2Fix).

BLASTp searches [52] against the NCBI conserved domain database using the translated

coding sequence of abra_1538 show it to have high similarity with the family of

periplasmic binding domains of basic membrane lipoproteins. The proteins within this

family are localized to the cell surface, and contain a sugar-binding protein-like fold,

which could be important in transport of EPS or LPS components. Members of this

family include Med, a positive transcriptional regulator from B. subtilis [175] and PnrA

from Treponema pallidum, a purine nucleoside binding receptor belonging to an ABC

transport system [176], suggesting that abra_1538 may be involved in nutrient transport

or binding. The function of NLPA lipoprotein is unknown, although studies with

pathogenic forms of E. coli have suggested that it plays a role in vesicle formation [177].

134

Further, it is suggested to be structurally distantly related to solute binding proteins

known to transfer solutes from the inner membrane for concentration within the

cytoplasm [178]. Taken together, this data suggests that abra_1538 and abra_4083 play

an important role in Sp7 nitrogen fixing cells.

Interestingly, the protein with the largest degree of up-regulation in Sp7 nitrogen

fixing cells (26-fold, Table 3.8) is a protein annotated as 4-phytase (abra_6441). Phytase

(phytate-3-phosphatase) catalyzes hydrolysis of phosphate groups from phytate, the main

storage form of phosphorous in cereal grasses and legumes [179]. Phosphorous stored in

the form of phytate is poorly utilized by the plant, making microbial phytase a key

enzyme in phosphorous metabolism. Studies have shown that plants are better able to

use phosphorous in low phosphorous soils when inoculated with phytase-secreting

microbes [179]. Although there are four different classes of phytases, studies with

microbial genomes have indicated that beta-propeller phytases are the most common type

found in soil microbes [180]. Beta propeller phytases possess six calcium ion binding

sites scattered throughout the protein structure but have no conserved binding motif.

Further, phytases belonging to this group share very little sequence identity (< 25%)

between them [180]. A BLASTp [52] search of the translated coding sequence of

abra_6441 against the NCBI non-redundant protein database indicates that it shows

greater than 47% similarity to a number of extracellular solute binding proteins from

Roseobacter sp. and Bordatella sp.. The 26-fold up-regulation in Sp7 nitrogen fixing

cells indicates that this protein is likely to be important in nitrogen fixation, perhaps in

increasing phosphorous availability to the cells or in making phosphorous available for

use in phospholipids that contribute to cell membrane remodeling. This protein is also

135

detected with 50% sequence coverage in Sp245 nitrogen fixing cultures, although it is not

detected in Sp245 non-nitrogen fixing cultures, making it impossible to determine if it is

up-regulated or simply expressed only in nitrogen fixing conditions in Sp245 strain.

Sp7 cells grown under nitrogen fixing conditions are smaller in size and rounder,

thus increasing the cell surface-to-volume ratio, indicating active remodeling of

morphology. They also accumulate PHB (poly-hydroxybutyrate) granules, indicating a

need to store carbon for later use. Earlier studies with both endophytic and rhizospheric

Azospirillum species showed that different strains exhibited different growth rate

responses to the same substrates. All strains produce large amounts of

exopolysaccharides (EPS) in early growth phases [158], with amount produced being

inversely proportional to the amount of ammonium available to the cell [149]. The stored

EPS is then metabolized for energy during nitrogen fixation, while intracellular carbon

stores such as PHB are used primarily for cell maintenance activities [158]. The changes

observed in morphology of Sp7 nitrogen fixing cells as well the changes observed in

carbon utilization during nitrogen fixation are consistent with the changes in proteome

expression profiles of these cells.

Conclusions

Nitrogen fixation results in predictable changes in proteome expression profiles

reflecting the basic metabolic shifts necessary for adaptation to nitrogen fixing

conditions. The slower growth observed in these cells is reflected in the proteome by

decreased numbers and amounts of ribosomal proteins, elongation factors, and DNA and

RNA binding proteins. The greater energy requirements for nitrogen fixation are found

within the proteome in the form of increased expression of electron transport components

136

to provide energy and increased expression of certain TCA cycle enzymes to provide

additional reducing power. The most notable change noted in the proteome is in the

greater abundance and variety of transport molecules, suggesting a greater need for

molecular transport in nitrogen fixing cells. This is especially noted in both number and

amount of those proteins expressed that are predicted to be involved in branched chain

amino acid transport. These transport components also appear quite diverse and they are

apparently not genetically linked, suggesting a global change in translation that affects

the entire physiology of the cell.

As stated earlier, Sp7 and Sp245 occupy different ecological niches, and thus

must adapt to different situations, which could be distinguished by different physiology

and metabolism between the strains when studied in vitro. Proteomic investigations have

revealed a number of similarities between these two species, with 959 proteins commonly

expressed in both Sp7 and Sp245 cultures grown under conditions of molybdate nitrogen

fixation. Proteins that are commonly up-regulated in all nitrogen fixing cultures (Table

3.10), are likely to play very important roles in adaptation to nitrogen fixing conditions.

This includes proteins such as solute binding proteins abra_0139 and 0141 and UspA

domain protein abra_3154 as mentioned earlier, but also includes a number of other

proteins involved in molecular transport as well as a number of hypothetical proteins.

Perhaps most interesting to note is the pattern of up- and down-regulation

between nitrogen fixing cultures of each species. In Table 3.10, a number of proteins are

not detected to be up-regulated in Sp7 cultures that show substantial fold change levels of

up-regulation in Sp245 nitrogen fixing cultures. This same pattern is noted with down-

137

regulated proteins as shown in Table 3.9. However, the majority of the up-regulated

proteins are detected in nitrogen fixing cultures (indicated by NF in Table 3.10), many

with sequence coverage of 50 % or greater, but are not detected in non-nitrogen fixing

Sp7 cultures. Therefore, a level of up-regulation could not be determined for these

proteins. For down-regulated proteins, the majority of those proteins were noted in Sp7

control (nitrogen replete) cells, most at low levels of sequence coverage, but were not

noted in nitrogen fixing cells, so a level of down-regulation could not be determined. This

could be due to the large numbers of ribosomal proteins in Sp7 nitrogen replete control

cultures which decreased the depth of proteome coverage and resulted in lower numbers

of proteins detected as discussed earlier. Overall, the data suggests that Sp7 species has

less dramatic changes in protein abundance than does the Sp245 species. This could,

however, be an artifact related to the current lack of complete genome sequence for Sp7.

It is possible that some of the proteins found to be expressed in the Sp245 proteome are

simply not present in the Sp7 genome. Another possibility is that many unknown

proteins found in Sp7 could have similar functions as in Sp245. This comparative

proteome study has nevertheless revealed differences in the physiology of cells under

different growth conditions. Further, a subset of proteins has been identified that may

distinguish the unique physiology of Sp7 and Sp245 under different growth conditions.

138

Chapter 4. Comparison of Proteome Expression between

Azospirillum brasilense wild-type and mutants in a chemotaxis-

like signal transduction pathway

Introduction

As discussed in Chapter 1, a number of bacteria have multiple chemotaxis

operons that can mediate response to a variety of environmental conditions [111].

A. brasilense species have a large genome size and exhibit extensive ability to adapt to a

number of different environmental conditions. The draft sequence of A. brasilense Sp245

indicates the presence of four chemotaxis-like operons and 45 sensory transducers. One

of these chemotaxis-like operons, named Che1 [117], has been previously characterized

[139] and found to contribute to chemotaxis, although that is not its primary function

[118]. The Che1 operon comprises components that are typically found in prototypical

chemotaxis pathways that regulate motility patterns. As discussed in Chapter 1, this

includes a two component signal transduction pathway consisting of a histidine kinase,

CheA1 and its cognate response regulator, CheY1, a coupling protein CheW1 and

adaptation proteins, chemoreceptor-specific CheB1 methylesterase and CheR1

methyltransferase. Although the operon structure suggests involvement in chemotaxis,

experimental evidence indicates that Che1 gene products function to modulate both cell

length at division and cell-to-cell aggregation (named clumping) as well as contribute to

chemotaxis [117]. In particular, mutants lacking a functional forward signaling pathway

due to deletion of genes encoding histidine kinase CheA1, response regulator CheY1 or

deleted for the entire operon, Che1, were found to be shorter in length than wild-type, to

maintain a smooth swimming direction for longer periods of time, and to flocculate

sooner and more than wild-type [117]. In contrast, a mutant lacking a functional

139

adaptation pathway (or “molecular memory”) due to deletion of genes encoding both

CheB1 methylesterase and CheR1 methyltransferase, here termed ∆cheB1cheR1, were

found to be longer than wild-type, to divide at longer lengths than wild type, to be

defective in chemotaxis and to be impaired (with substantial delays) in flocculation [117].

Both mutants were found to make less exopolysaccharide (EPS) than wild-type, with the

colonies of the ∆cheY1 mutant having a wrinkled, wet appearance distinctly different

from wild-type, while the colony morphology of the ∆cheB1cheR1 mutant was similar to

wild-type [117]. From the above experimental results, it can be hypothesized that the

Che1 operon has an indirect effect on chemotaxis with its primary function being the

control of cell length at division and the propensity for cell-to-cell interaction (clumping).

However, as stated above, the molecular components of the Che1 pathway show

extensive similarity to that found in prototypical chemotaxis signal transduction pathways

that exclusively function in controlling the swimming motility pattern. The molecular

basis of the Che1 control of cell length at division and clumping remains to be

determined. In order to identify potential pathways and functions that may be regulated

by this operon as well as possible physiological differences that result from the activity of

this operon, we here undertake a proteomic comparison of wild type Sp7 cultures versus

mutant ∆cheB1cheR1 and mutant ∆cheY1 derivative strains cultures.

Materials and methods

Bacterial Strains and Cell growth

Wild–type A. brasilense strain Sp7 (ATCC 29145) was used as control throughout

this study, and is here referred to as Sp7. The mutant lacking functional CheB1 and

140

CheR1 [∆(cheB1-cheR1)::Km] (BS104) was constructed previously [118], and is herein

referred to as ∆cheB1cheR1. The mutant lacking functional CheY1 [∆(CheY1)::Km]

(AB102) was also constructed earlier [117] and is referred to in this work as ∆cheY1.

All cultures were grown at 25°C on an orbital shaker in minimal media (MMAB)

[139] with 5 mM malate and 5 mM fructose as carbon sources, and ammonium chloride

as nitrogen source. One culture was grown for proteome preparations. No replicate was

grown, so all studies include only one culture. Because earlier studies indicated a

maximum difference in size at low optical density (below OD600 0.5), cultures were

harvested at an OD600 0.3, and cell pellets weighed and frozen at -80°C for later proteome

preparation. A crude membrane fraction for the Sp7 control was prepared as follows

from a culture grown to an OD600 of 0.6. Cells were harvested and lysed by sonication.

Cleared lysate was centrifuged at 18000 xg in a Sorvall ultracentrifuge, and the pellet

resolubilized in 50 mM Tris-HCl, pH 7.5 to a protein concentration of 1 mg/ml.

Proteome preparation

Proteome preparation was done as discussed earlier. Briefly, cell pellets were re-

suspended at a rate of 0.1g wet cell pellet weight per 500 µl buffer in 6M guanidine

hydrochloride and lysed by sonication. Lysate was then cleared of debris by

centrifugation at 12000 xg in a Sorvall centrifuge, and cleared lysate was diluted for

subsequent digestion. After addition of 10 mM DTT and 1 hour incubation at 60°C,

sequencing-grade trypsin (Promega) was added, and solution was incubated overnight at

37°C. An additional 20 µg trypsin was added to each sample the following morning, and

samples were incubated for an additional 6 hours. Digestion was halted by addition of

141

0.1% formic acid. Digested samples were then desalted using Sep-Pak Plus C-18 solid

phase extraction (Waters, Milford, MA) following manufacturer’s recommendations.

Eluates were solvent exchanged via vacuum centrifugation into HPLC-grade water

(Burdick-Jackson) containing 0.1% formic acid. Samples were aliquoted into 40 µl

volumes and frozen at -80°C for later analysis.

LC/LC-MS/MS

Analysis of cleaned, digested samples was accomplished using MudPIT [17, 20]

as described in previous chapters, with each sample being analyzed in technical

duplicates. Only two technical replicates were run due to limited instrument time

available. Briefly, samples were loaded via pressure cell onto a back column consisting

of 3 cm reverse-phase (RP) material followed by 3 cm strong cation exchange (SCX).

This back column was then coupled to a 15 cm RP resolving column, packed into a fused

silica column pulled to a tip with a 5 µm opening. This front column was then placed

directly in-line with a linear ion trapping quadrupole mass spectrometer (LTQ, Thermo-

Finnagan). Chromatography was accomplished in 12 steps, with each step consisting of

an initial salt pulse to elute a subset of peptides from SCX, followed by a gradient elution

of these subsets of peptides with increasing concentration of organic solvent, as described

more thoroughly in earlier chapters. Peptides were eluted directly into the mass

spectrometer via nano-electrospray for analysis.

The mass spectrometer was operated according to parameters established in

Chapter 2 of this work, with instrument operation parameters and data collection being

142

under the control of Xcaliber software. Data was collected in data dependent mode, with

one parent ion scan being followed by six tandem MS scans.

Data Analysis

Raw files were searched using SEQUEST [3] against the Sp245 database created

earlier as described in Chapter 3 since an Sp7 species-specific database is not yet

available. Search parameters were as follows: tryptic digestion, peptide mass tolerance of

3 m/z and a fragment ion tolerance of 0.5 m/z. Additionally, search parameters included

two dynamic modifications: 1. methylation represented by a mass shift of +14 m/z on

glutamate residues, and 2. deamidation followed by methylation represented by a mass

shift of +15m/z on glutamine residues in order to facilitate later analysis of receptor

methylation patterns which were beyond the scope of this dissertation. Results were then

filtered and collated using DTASelect [54] with the following parameters: XCorr filter

levels of 1.8 for peptides with a charge state of +1, 2.5 for those with charge state +2 and

3.5 for charge state +3, minimum delta CN of 0.08, non-tryptic status and 2 peptides per

protein identification. Normalized spectral abundance factors (NSAF) [72] were

determined for each protein in a given run, and were added to the search results. All

DTASelect output files, both HTML and text formats, can be found at

https://compbio.ornl.gov/mspipeline/ azospirillum / study1/ status.html.

Search results were imported into Microsoft Access for analysis. Two replicates

of each sample were used in all data analysis, and comparisons were done using Sp7

wild-type cultures grown to an OD600 0.3 as control. Pie charts were constructed using

the total percentages of proteins expressed in each individual functional category based

143

on ΣNSAF values of each individual protein expressed. Levels of up-regulation and

down-regulation were determined as described in Chapter 3, with NSAF values of each

protein in mutant cultures divided by NSAF values of the same protein in Sp7 control

cultures. Although potentially interesting, proteins expressed only in mutant cultures

were not considered in this analysis since levels of up- or down-regulation could not be

determined. Further, filter levels of 2.5-fold change were applied to these data sets

because the Sp7 nitrogen fixing data sets examined in Chapter 3 of this work indicated a

lower level of up- or down-regulation in Sp7 strain. Additionally, because work with

individual Sp7 genes located on the pRHICO plasmid have suggested a possibility that

some genes, and perhaps their gene products, can vary significantly from the sequence

found in the Sp245 genome [172], a BLASTx [134] comparison of available Sp7

translated coding sequences against the Sp245 genome was done.

Results

Sixty seven individual gene sequences for Azospirillum brasilense strain Sp7 are

available in the NCBI database (http://www.ncbi.nlm.nih.gov/). BLASTx [134] searches

using translated coding sequences from these Sp7 strain sequences indicate that 95% of

these genes show between 90% and 100% similarity to coding sequences in the Sp245

genome. The high degrees of similarity between coding sequences of the Sp7 genes and

those of the Sp245 genome lend credibility to those protein identifications made using the

Sp245 genome. One Sp7 translated coding sequence, from gene nirS, does not have a

homolog in the Sp245 genome. Two Sp7 translated coding sequences, molybdenum

nitrogenase alpha and beta subunits, show a low degree of similarity (27% and 32%,

respectively) to those from the Sp245 genome, as discussed in Chapter 3. While some

144

proteins may not be identified in the Sp7 proteome by using the translated coding

sequences from the Sp245 genome, the high degree of similarity of available Sp7

translated coding sequences to the Sp245 genome sequences gives confidence to those

identifications that are made.

Overall proteome analysis

The Azospirillum brasilense Sp7 control (Sp7) yielded 1366 identifications from

combined technical replicate runs, while the ∆cheB1cheR1 and ∆cheY1 mutant strains

gave 1355 and 1280 identifications, respectively. 921 expressed proteins were common

to all cultures, and reproducibility (determined by the number of common proteins

detected in both runs (not shown)/ total number of proteins detected in combined

datasets) between technical replicates was 67% for ∆cheY1 samples and 71% for both ∆

cheB1cheR1 and Sp7 cultures (Table 4.1). ∆cheY1 and ∆cheB1cheR1 cultures had 1003

common protein identifications, giving a good representation of proteins required for

basic metabolic functions in both mutant cultures. Attached spreadsheet

Mutant_Proteome_Data gives a list of all proteins detected in each culture.

The assigned protein identifications were grouped according to functional

category as described in Chapter 3, and the numbers of proteins identified in each

category for each sample are given in Table 4.2. Also as described in Chapter 3, the

relative abundance of each protein detected within a sample was determined through

calculating a normalized spectral abundance factor (NSAF) for each protein identified

[181]. NSAF values were summed for all proteins in each individual functional category

(ΣNSAF) and the percentage of totaled ΣNSAF values of identified proteins that belong

145

Table 4. 1 Total number of protein identifications in each Sp 7 technical replicate run

Number of Protein Identifications Reproducibility between runs

Sample Replicate 1 Replicate 2 Combined Dataset

Sp7 OD 0.3 1179 1160 1367 71.1%

Sp7 ∆cheY1 mutant 1077 1074 1281 67.0%

Sp7 ∆cheB1cheR1 mutant 1087 1237 1356 71.3%

146

Table 4. 2 Comparison of protein expression in the cultures of the parental strain Sp7 and its ∆∆∆∆cheB1cheR1 and ∆∆∆∆cheY1 mutant derivatives by

functional category

Sp7 OD 0.3 ∆cheB1cheR1 ∆cheY1

FC Functional Category Description No. Common

Proteinsa

No.

Proteinsb

ΣNSAF

Percentagec

No.

Proteinsb

ΣNSAF

Percentagec

No.

Proteinsb

ΣNSAF

Percentagec

B Chromatin structure and dynamics 0 1 NA 1 NA 1 NA

C Energy production and Conversion 89 118 11.6% 114 12.5% 104 10.4%

D Cell Division and Chromosome Partitioning 16 20 0.5% 20 0.6% 17 0.7%

E Amino Acid Transport and Metabolism 120 157 8.7% 152 8.9% 143 8.3%

F Nucleotide Transport and Metabolism 37 52 2.9% 47 3.2% 44 2.9%

G Carbohydrate Transport and Metabolism 41 60 4.8% 58 5.2% 51 4.3%

H Coenzyme Metabolism 38 71 1.9% 67 2.0% 50 1.8%

I Lipid Metabolism 28 44 1.5% 39 1.8% 39 1.9%

J Translation, Ribosomal Structure and Biogenesis 118 133 32.2% 138 29.2% 121 29.3%

K Transcription 35 52 3.9% 55 3.8% 49 3.7%

L DNA Replication, Recombination and Repair 26 46 2.3% 39 2.3% 44 2.7%

M Cell Envelope Biogenesis, Outer Membrane 40 55 3.7% 60 3.5% 61 4.1%

N Cell Motility and Secretion 12 33 0.9% 40 0.6% 41 0.7%

O Posttranslational Modification, Protein turnover,

Chaperones

56 69 6.9% 63 6.3% 69 7.3%

P Inorganic Ion Transport and Metabolism 34 42 2.9% 42 3.7% 42 2.6%

Q Secondary metabolites biosynthesis, transport

and catabolism

17 23 0.9% 24 0.9% 22 1.5%

R General Function Prediction Only 59 98 2.8% 87 2.7% 87 2.7%

S Function Unknown 42 78 2.7% 76 2.8% 71 3.4%

T Signal Transduction Mechanisms 26 51 1.2% 59 1.7% 54 1.6%

U Intracellular trafficking and secretion 10 10 0.3% 13 0.4% 11 0.5%

V Defense mechanisms 1 4 0.05% 3 0.03% 10 0.1%

AA Hypotheticals with similarity or homology to known 34 60 3.5% 64 4.2% 56 5.2%

BB Hypothetical 42 90 3.5% 95 3.8% 94 4.2%

TOTAL 921 1366 99.6% 1355 100% 1280 100% a Number of proteins identified in every growth state b Number of proteins identified in each functional category in that growth state c NSAF values were totaled for proteins in each individual functional category(ΣNSAF) to determine the percentage of the total expressed proteome within each functional

category

147

to each category were calculated based on their overall abundance as represented by

NSAF values (Table 4.2). A graphical representation of proteome expression profiles

compiled by percentage of ΣNSAF values totaled for each functional category is given in

Figure 4.1. The proteome expression profiles for all cultures look remarkably similar

given the differences in length and colony morphology noted in the different mutant cells

[117].

Differences between mutant cultures are more apparent in the expression levels of

individual proteins. Levels of up-regulation and down-regulation in comparison to

control cultures were determined as described in Chapter 3, dividing the calculated NSAF

value of the protein in a mutant culture by the calculated NSAF value of the same protein

in a control Sp7 wild-type culture to determine levels of up-regulation, and dividing

calculated NSAF values of the Sp7 control divided by those of the same protein in a

mutant culture to determine levels of down-regulation. Table 4.3 shows the levels of up-

regulation (fold-change) of proteins in both mutant cultures when compared to Sp7

controls, while Table 4.4 shows the levels of down-regulation (fold-change) of proteins in

both mutant cultures when compared to Sp7 controls. Examination of these tables clearly

illustrates a very different pattern of differential protein expression levels in ∆cheY1

mutants versus ∆cheB1cheR1 mutants when compared to Sp7 controls, discussed more

thoroughly below.

Perhaps most intriguing in the ∆cheY1 proteome is upregulation of abra_4441

type VI secretion protein. Although this protein is detected in Sp7 wild-type controls, it

is not detected in ∆cheB1cheR1 mutants at all. Type VI secretion systems (T6SS) are

148

Figure 4. 1 Proteome expression profiles for A) wild-type Sp7 control, B) mutants lacking functional CheB1 and CheR1, and C) mutants lacking

functional CheY1

Proteins expressed are grouped by functional category and NSAF values for each individual protein within that category are totaled (ΣNSAF). From the

total ΣNSAF of each category, a total percentage of proteins in each functional category based on relative abundance as represented by NSAF values is

calculated. Pie chart representations indicate that all cultures are growing rapidly, with only slight differences in the protein expression profile of the

mutants versus the wild-type control.

CheY1 mutant

AA

5% BB

4%

C

10%

E

8%

F

3%

G

4%

H

2%

I

2%

J

29%

K

4%

L

3%

M

4%

O

7%

P

3%

R

3%

S

3%

T

2%

Sp7 OD 0.3

AA

4%BB

4%

C

12%

E

9%

F

3%

G

5%

H

2%

I

2%

J

32%

K

4%

L

2%

M

4%

O

7%

P

3%

R

3%

S

3%

CheB1CheR1 mutant

AA

4%BB

4%

C

12%

E

9%

F

3%

G

5%

H

2%

I

2%

J

29%

K

4%

L

2%

M

3%

O

6%

P

4%

R

3%

S

3%

T

2%

A)Wild-type Sp7 B)∆cheB1cheR1 C)∆cheY1

149

Table 4. 3 Fold change of proteins up-regulated in both ∆cheB1cheR1 and ∆cheY1 mutants versus Sp7 controls

Gene Product FuncCat

Avg fold-

change upreg ∆cheB1cheR1 ∆cheY1

abra_0714 NUDIX hydrolase AA 2.7 2.7

abra_1169 phasin family protein AA 5.0 2.9 7.1

abra_1276 ribosomal protein L36 AA 2.8 2.8

abra_2407 flagellar assembly regulator FliX AA 2.6 2.6

abra_3170 phasin family protein AA 2.9 2.6 3.3

abra_6288 histidine kinase HAMP region domain protein AA 5.2 5.2

abra_1987 conserved hypothetical protein BB 4.3 4.3

abra_2391 conserved hypothetical protein BB 3.5 3.5

abra_2655 hypothetical protein BB 5.2 5.2

abra_3281 Sporulation domain protein BB 3.0 3.0

abra_3348 hypothetical protein BB 4.1 4.1

abra_3990 hypothetical protein BB 13.3 8.5 18.0

abra_4002 hypothetical protein BB 3.8 3.8

abra_4507 flagellin domain protein BB 2.6 2.6

abra_6429 conserved hypothetical protein BB 4.0 4.0

abra_0437 iron-sulfur cluster binding protein C 3.7 3.7

abra_2275 cytochrome c oxidase, subunit II C 3.5 3.5

abra_2759 cytochrome c prime C 2.9 2.9

abra_4337 Cytochrome b/b6 domain C 2.6 2.6

abra_4829 ubiquinol oxidase, subunit II C 2.7 2.7

abra_4904 Pyridoxal 4-dehydrogenase C 2.8 2.8

abra_6420 Phosphoenolpyruvate carboxylase C 2.9 2.9

abra_0574 cell divisionFtsK/SpoIIIE D 3.5 3.5

abra_1289 Cobyrinic acid ac-diamide synthase D 3.0 3.0

abra_1445 TonB box domain protein D 2.8 2.8

abra_4240 chromosome segregation protein SMC D 2.7 2.7

abra_5418 Cobyrinic acid ac-diamide synthase D 4.3 3.6 5.0

abra_0573 aminotransferase class I and II E 2.6 2.6

abra_0897 Prephenate dehydratase E 2.5 2.5

abra_1479 Extracellular ligand-binding receptor E 4.1 4.1

abra_1565 nitrogen regulatory protein P-II E 5.3 5.3

150

Gene Product FuncCat

Avg fold-

change upreg ∆cheB1cheR1 ∆cheY1

abra_1858 Pyridoxal-5'-phosphate-dependent protein beta subunit E 3.5 2.8 4.2

abra_1864 aminotransferase class I and II E 3.0 3.0

abra_1935 spermidine/putrescine ABC transporter ATPase subunit E 2.8 2.8

abra_2184 phosphoadenosine phosphosulfate reductase E 2.7 2.7

abra_2247 Glycine dehydrogenase (decarboxylating) E 3.2 3.2

abra_3996 ornithine carbamoyltransferase E 4.4 4.4

abra_4110 4-aminobutyrate aminotransferase E 2.9 2.9

abra_4286 Prephenate dehydrogenase E 3.1 2.9 3.4

abra_2870 Dihydroorotate oxidase F 4.0 4.0

abra_0068 Phosphomannomutase G 2.6 2.6

abra_3118 glycogen/starch/alpha-glucan phosphorylase G 3.5 3.5

abra_4739 KDPG and KHG aldolase G 2.7 2.7

abra_6513 glucose sorbosone dehydrogenase G 3.2 3.2

abra_6523 TRAP dicarboxylate transporter, DctP subunit G 3.8 4.1 3.5

abra_0548 glutathione synthetase H 3.5 3.5

abra_0553 oxygen-independent coproporphyrinogen III oxidase H 2.7 2.7

abra_1554 phosphomethylpyrimidine kinase H 4.2 4.2

abra_1954 5-formyltetrahydrofolate cyclo-ligase H 2.6 2.6

abra_2162 riboflavin synthase, alpha subunit H 3.5 3.5

abra_3240 pantetheine-phosphate adenylyltransferase H 2.6 2.6

abra_2604 6-phosphogluconate dehydrogenase, NAD-binding I 4.0 4.0

abra_3050 2-C-methyl-D-erythritol 4-phosphate cytidylyltransferase I 2.8 2.8

abra_3442 phosphatidyl-N-methylethanolamine N-methyltransferase I 3.6 3.6

abra_5859 acetyl-CoA acetyltransferase I 3.0 3.0

abra_0826 tRNA (guanine-N1)-methyltransferase J 3.2 3.2

abra_2439 ribosomal protein L24 J 2.7 2.7

abra_2440 ribosomal protein L14 J 4.2 4.2

abra_2443 ribosomal protein L16 J 3.2 3.2

abra_6528 GntR domain protein K 3.2 3.2

abra_1522 AAA ATPase central domain protein L 2.7 2.7

abra_1817 replicative DNA helicase L 2.5 2.5

Table 4.3 (cont) Fold change of proteins upregulated in both ∆cheB1cheR1 and ∆cheY1 mutants versus Sp7 controls

151

Gene Product FuncCat

Avg fold-

change upreg ∆cheB1cheR1 ∆cheY1

abra_2111 transcription-repair coupling factor L 4.2 4.2

abra_1829 Lytic transglycosylase catalytic M 2.9 2.9

abra_1965 peptidoglycan-associated lipoprotein M 3.0 2.7 3.4

abra_1966 polysaccharide biosynthesis protein CapD M 2.6 2.6

abra_2091

glucosamine/fructose-6-phosphate aminotransferase,

isomerizing M 2.8 2.8

abra_6505 efflux transporter, RND family, MFP subunit M 11.3 11.3

abra_2408 flagellar P-ring protein N 3.1 3.1

abra_2567 chemotaxis sensory transducer N 5.4 2.8 8.0

abra_3025 chemotaxis sensory transducer N 3.1 3.1

abra_5341 chemotaxis sensory transducer N 3.2 3.2

abra_5409 PAS sensor protein N 2.6 2.6

abra_0257 cytochrome c-type biogenesis protein CcmI O 2.7 2.6 2.8

abra_3997 Hsp33 protein O 2.9 2.9

abra_4742 ATP-dependent chaperone ClpB O 4.2 4.2

abra_1378 Carbonate dehydratase P 3.4 3.4

abra_2177 potassium efflux system protein P 3.4 2.6 4.2

abra_2189 ABC transporter related P 2.8 2.8

abra_0652 Carboxymethylenebutenolidase Q 3.8 3.8

abra_3360 5-oxopent-3-ene-1,2,5-tricarboxylate decarboxylase Q 2.8 2.9 2.7

abra_5423 acetoacetyl-CoA reductase Q 2.9 2.9

abra_6359 acetoacetyl-CoA reductase Q 3.9 3.9

abra_0487 alanine racemase domain protein R 2.6 2.6

abra_1766 enoyl-(acyl carrier protein) reductase R 4.1 4.1

abra_1884 TRAP transporter solute receptor, TAXI family R 3.5 3.5

abra_2263 Mitochondrial processing peptidase R 2.5 2.5

abra_3266 protein of unknown function DUF815 R 3.3 3.3

abra_4621 Serine/threonine protein kinase-related R 3.0 3.0

abra_6620 basic membrane lipoprotein R 2.6 2.5 2.7

abra_0139 conserved hypothetical protein S 30.3 34.2 26.5

abra_0151 Heparinase II/III family protein S 2.9 2.9

Table 4.3 (cont) Fold change of proteins upregulated in both ∆cheB1cheR1 and ∆cheY1 mutants versus Sp7 controls

152

Gene Product FuncCat

Avg fold-

change upreg ∆cheB1cheR1 ∆cheY1

abra_0162 HemY domain protein S 2.9 2.9

abra_1380 protein of unknown function DUF1013 S 2.9 2.9

abra_1388 protein of unknown function DUF526 S 3.9 3.9

abra_2142 protein of unknown function DUF490 S 6.5 6.5

abra_2717 40-residue YVTN family beta-propeller repeat protein S 2.7 2.7

abra_2767 virulence protein SrfB S 2.7 2.7 2.7

abra_2768 conserved virulence factor protein S 3.9 3.0 4.8

abra_3284 iron-sulfur cluster assembly accessory protein S 3.9 3.9

abra_4441 type VI secretion protein, VC_A0107 family S 2.8 2.8

abra_4455 type VI secretion-associated protein, ImpA family S 2.6 2.6

abra_4533 putative periplasmic ligand-binding sensor protein S 2.9 2.9

abra_0407 PhoH family protein T 3.2 2.7 3.7

abra_0805 response regulator receiver T 3.5 3.5

abra_2692 response regulator receiver T 3.0 3.0

abra_3691 PAS sensor protein T 3.6 3.6

abra_4542 Stage II sporulation E family protein T 3.5 2.8 4.2

abra_5331 PAS sensor protein T 3.3 3.3

abra_3264 preprotein translocase, YajC subunit U 2.7 2.8 2.6

Table 4.3 (cont) Fold change of proteins upregulated in both ∆cheB1cheR1and ∆cheY1 mutants versus Sp7 controls

153

Table 4. 4 Fold change of proteins down-regulated in both ∆cheB1cheR1 and ∆cheY1 versus Sp7 controls

Gene Product FuncCat

Avg fold-change

Downreg ∆cheB1cheR1 ∆cheY1

abra_0403 helix-turn-helix domain protein AA 6.5 3.5 9.5

abra_3900 NLP/P60 protein AA 2.5 2.5

abra_3938 acyl-CoA reductase AA 2.5 2.5

abra_0735 conserved hypothetical protein BB 4.6 4.6

abra_0844 Hpt domain protein BB 2.6 2.6

abra_2512 conserved hypothetical protein BB 2.6 2.6

abra_2736 hypothetical protein BB 2.7 2.7

abra_2750 conserved hypothetical protein BB 2.6 2.6

abra_0918 dihydrolipoamide dehydrogenase C 2.7 2.7

abra_1181 molybdopterin oxidoreductase C 10.4 17.8 3.0

abra_1305 ferredoxin C 4.0 4.0

abra_1578 NADH-quinone oxidoreductase, F subunit C 3.6 3.6

abra_3380 D-isomer specific 2-hydroxyacid dehydrogenase NAD-binding C 2.6 2.6

abra_0517 spermidine synthase E 3.5 3.5

abra_0979 2-isopropylmalate synthase/homocitrate synthase family protein E 3.1 3.1

abra_1757 urease, gamma subunit E 3.0 3.0

abra_1865 ABC transporter related E 3.1 3.1

abra_1868 cationic amino acid ABC transporter, periplasmic binding protein E 3.4 3.4

abra_2642 peptidase M24 E 2.7 2.7

abra_2966 aspartate-semialdehyde dehydrogenase E 3.6 3.6

abra_3967 aspartate kinase E 2.8 2.8

abra_4885 Substrate-binding region of ABC-type glycine betaine transport system E 3.2 3.2

abra_4887 glycine betaine/L-proline ABC transporter, ATPase subunit E 3.4 3.4

abra_1146 Orotate phosphoribosyltransferase F 4.3 4.3

abra_6433 thiazole biosynthesis family protein F 2.7 2.7

abra_0423 PfkB domain protein G 2.8 2.8

abra_0921 transaldolase G 3.7 3.7

abra_1553 phosphoglucosamine mutase G 2.8 2.7 2.9

abra_1642 Triose-phosphate isomerase G 2.7 2.7

abra_1950 glyceraldehyde 3-phosphate dehydrogenase G 3.3 3.3

abra_2097 pantoate/beta-alanine ligase H 2.6 2.6

154

Gene Product FuncCat

Avg fold-change

Downreg ∆cheB1cheR1 ∆cheY1

abra_2597 Polyprenyl synthetase H 2.7 2.7

abra_4347 deoxyxylulose-5-phosphate synthase H 2.9 2.9

abra_0946 ribosomal protein L27 J 2.8 2.7 3.0

abra_0980 RNA methyltransferase, TrmH family, group 1 J 4.3 4.3

abra_2435 ribosomal protein L6 J 3.1 3.1

abra_3254 gid protein J 2.6 2.6

abra_3690 tyrosyl-tRNA synthetase J 2.9 2.9

abra_2463 transcription termination/antitermination factor NusG K 3.1 3.1

abra_6192 LysR substrate-binding K 3.9 3.9

abra_4247 DNA methylase N-4/N-6 domain protein L 2.5 2.5

abra_3497 Serine-type D-Ala-D-Ala carboxypeptidase M 3.2 3.2

abra_2635 chemotaxis sensory transducer N 3.3 3.3

abra_6213 flagellin domain protein N 15.0 7.0 23.0

abra_0172 2-alkenal reductase O 2.7 2.7

abra_0444 20S proteasome A and B subunits O 4.3 4.3

abra_0471 protein of unknown function DUF255 O 4.8 4.8

abra_4609 Endopeptidase Clp O 2.9 2.9

abra_3260 Superoxide dismutase P 2.9 2.9

abra_6242 protein of unknown function DUF47 P 2.5 2.5

abra_6590 conserved hypothetical signal peptide protein P 2.6 2.6

abra_0566 ABC transporter related Q 3.0 3.0

abra_0593 ThiJ/PfpI domain protein R 6.6 6.6

abra_0945 GTP-binding protein Obg/CgtA R 3.0 3.0

abra_1538 basic membrane lipoprotein R 2.6 2.6

abra_3520 cobalamin biosynthesis protein CobW R 5.6 5.6

abra_0131 protein of unknown function DUF971 S 2.9 2.9

abra_0464 protein of unknown function DUF299 S 2.5 2.5

abra_1433 rpsU-divergently transcribed protein S 3.0 3.0

abra_0303 CBS domain containing protein T 3.6 3.6

Table 4.4 (cont) Fold change of proteins downregulated in both ∆cheB1cheR1 and ∆cheY1

versus Sp7 controls

155

Table 4. 5 NSAF values of Type 6 Secretion System components detected in mutant strains and in Sp7 control cultures

Type 4 Secretion System IcmF, DotU region

∆cheB1cheR1 ∆cheY1 Sp7

abra_0639 hypothetical protein 8E-4 7E-4 9E-4

abra_0643 dctP TRAP dicarboxylate transporter, DctP subunit 3E-4 3E-4 3E-4

abra_0646 type VI secretion protein IcmF 5E-5 6E-5 NDa

abra_0647 type IV / VI secretion system protein, DotU family NDa ND

a 5E-5

Type 6 Secretion System region

∆cheB1cheR1 ∆cheY1 Sp7

abra_4428 sigma-54 factor interaction domain-containing protein 6E-5 NDa 6E-5

abra_4435 type VI secretion ATPase, ClpV1 family NDa 2E-4 8E-5

abra_4439 protein of unknown function DUF796 NDa 5E-4 ND

a

abra_4440 type VI secretion protein, EvpB/VC_A0108 family NDa 7E-4 5E-4

abra_4441 type VI secretion protein, VC_A0107 family NDa 2E-3 8E-4

abra_4442 conserved hypothetical protein NDa 2E-4 1E-4

abra_4452 type VI secretion protein, VC_A0114 family 5E-5 NDa 1E-4

abra_4455 type VI secretion-associated protein, ImpA family NDa 2E-3 1E-5

abra_4458 hypothetical protein NDa 1E-4 ND

a

aND = not detected

156

multi-component transport systems, thought to be involved in bacterial interaction with a

eukaryotic host. Table 4.5 catalogs all components of the T6SS expressed in all cultures

with their concomitant NSAF values. Additionally, the NSAF values of Type 4 secretion

system that are homologs to the T6SS components are given in Table 4.5. While

components of the T4SS are detected in all cultures, several components of the T6SS are

detected in both ∆cheY1 mutant proteomes and in Sp7 wild-type control cultures, but are

not seen in the ∆cheB1cheR1 mutant at all, suggesting the intriguing possibility that this

system is responding, either directly or indirectly, to the Che1 signaling system.

Discussion

Proteome expression profiles are similar for control and mutants with only slight

differences noted. Since all cultures were harvested at similar points in their growth

(early log phase) and all were growing rapidly and dividing at the same rate, this result is

not unexpected and it is consistent with previous observations that indicated that

differences between the mutants and the parental strain were subtle [117]. The slight

differences noticed in percentage of abundances as represented by total NSAF values for

proteins in individual functional categories could be attributed to general variation in

culture growth.

As expected for early mid-log phase cultures, the most abundant proteins found

within all cultures grown (attached spreadsheet Mutant_Proteome_Data) indicate the

cultures were growing rapidly and were metabolically active. Active protein synthesis is

suggested by the abundance and variety of ribosomal proteins, chaperonins and

translation elongation factors, a common set of which is found in both mutant cultures as

157

well as in the Sp7 wild-type control culture. DNA manipulation and RNA synthesis is

implied by the presence of a number of DNA binding proteins. All cultures have an

abundance of ATP synthase subunits, cytochrome c proteins and electron transferring

flavoproteins, suggesting they are actively respiring. Several malate dehydrogenase

proteins and glyceraldehyde 3-phosphate dehydrogenase proteins are present in all

cultures, suggesting active glycolysis and TCA cycles as expected in early mid-log phase

cultures.

Differential expression in proteins related to general metabolism

Patterns of up- and down-regulation (Tables 4.3 and 4.4, respectively) are very

different for ∆cheB1cheR1 and ∆cheY1 mutants, indicating that each mutant has a unique

physiological response to absence of different components of the Che1 pathway. As

noted in other bacterial species, the lack of a forward signaling response regulator in a

two component signaling system consisting of histidine kinase-response regulator pair

elicits a more dramatic response from the mutant cells than those lacking an adaptation

pathway [182]. In keeping with this observation, ∆cheY1 mutants exhibit a greater

response to the loss of the forward signaling pathway than do ∆cheB1cheR1 mutants to

the loss of the adaptation pathway.

Protein metabolism

Protein metabolism appears to be very different in the two mutants. In ∆cheY1

mutants, a number of proteins involved in amino acid and cofactor biosynthesis as well as

in amino acid modification are found to be up-regulated, while only a few of these same

proteins show up-regulation in the ∆cheB1cheR1 mutant, suggesting an increased need

158

for amino acid metabolism in the ∆cheY1 mutant. Of specific interest are two proteins

involved in tyrosine metabolism, prephenate dehydratase (abra_0897) and prephenate

dehydrogenase (abra_4286), up-regulated in ∆cheY1 mutants by 2.6- and 3.4-fold,

respectively. Tyrosine residues have been found to be important in transport of

molecules through outer membrane protein OMP85 [174]. In E coli, OMP85 is the outer

membrane protein responsible for transport of a number of outer membrane proteins

[174]. OMP85 exists in a large complex in the outer membrane anchored to the

peptidoglycan by lipoprotein and recognizes a motif containing tyrosine residues

followed by three hydrophobic residues in those proteins it is to transport [174, 183].

Additionally, abra_4742 ClpB protease, an ATP-ase that functions to disaggregate

proteins and restore their function [184], is also up-regulated in ∆cheY1. Further,

glycine/betaine transporters, abra_4885 and 4887 are down-regulated by 3-fold in

∆cheY1 mutants. Betaines are zwitterionic compounds containing cationic functional

groups such as ammonium on one end of the molecule and anionic functional groups

such as carboxylate at the other end [185]. Glycine betaines are amino acid derivatives

that often serve as methyl donors, or can act as an osmolyte that both stabilizes proteins

and adjusts solute concentration in the cytoplasm to restore hydration and protect a cell

from dehydration [185]. Transporters act as osmolarity sensors in the surrounding

environment [185]. Down-regulation of these transporters may therefore simply be due

to the smaller size of the cells, but may also lead to a reduced ability to sense changes in

osmolarity of the surrounding media. Further, since glycine betaines serve to stabilize

proteins within the cell [185] the down-regulation of this protein in ∆cheY1 mutants could

lead to more aggregated proteins due to unfolding, thus leading to a higher expression of

159

ClpB protease as mentioned above. In contrast, ∆cheB1cheR1 mutants show few

proteins participating in amino acid metabolism to be up-regulated, instead exhibiting the

greatest degree of up-regulation in hypothetical proteins.

Molecular transport

Molecular transport appears to be different between the two mutants as well.

Interestingly, although ∆cheB1cheR1 mutants exhibit a larger cell size, ABC transporters

abra_1865 and abra_1868, as well as ABC transporter related protein abra_0566 are

down-regulated while only one ABC transporter, abra_2189, is up-regulated. In contrast,

∆cheY1 mutants do not show changes in expression levels of ABC transporter proteins.

However, both mutants show up-regulation of abra_3264 YajC preprotein translocase, a

component of Type III protein secretion systems which serves to translocate proteins

from the cytoplasm to the periplasmic space, suggesting the need for increased protein

translocation in both mutants. While this is understandable in the ∆cheB1cheR1 mutant

due to its larger size which presumably has more proteins on its outer membrane due to

the greater surface area of the membrane, it is an interesting phenomenon in the smaller

∆cheY1 cells.

Intriguingly, a protein found to be abundantly up-regulated in Sp245 cells grown

under nitrogen fixation, abra_0139, is also found to be among the most abundant proteins

of both ∆cheB1cheR1 and ∆cheY1 mutant proteomes. As discussed in Chapter 3,

abra_0139 is a small extracellular solute binding protein with a Bordatella Uptake Gene

(BUG) domain, thought to be involved in nutrient uptake. Interestingly, Sp7 cells grown

under nitrogen fixing conditions show a 9-fold up-regulation of abra_0139 (Chapter 3),

160

while the ∆cheB1cheR1 and ∆cheY1 mutants show a 34- and 26-fold up-regulation of this

protein, respectively (Table 4.3). An additional solute binding protein containing a BUG

domain, abra_0141, is also present in both mutant proteomes, but was not detected in

control cultures so levels of up-regulation could not be determined (attached spreadsheet

Mutant_Proteome_Data). This was also the case for Sp7 nitrogen fixing cultures where

abra_0141 was detected only in nitrogen fixing cells. The large degree of up-regulation of

solute binding receptor protein abra_0139 and the presence of solute binding receptor

protein abra_0141 in both mutant proteomes once again suggests that it is important in

nutrient uptake or transport, further suggesting that the ∆cheB1cheR1 and ∆cheY1

mutants are sensing their environment in a different way than are wild-type cells.

Carbon metabolism

Carbon metabolism also seems to be affected in the ∆cheY1 mutant. Evidence of

an altered carbon metabolism is given in the up-regulation of several proteins involved in

carbon metabolism. Abra_4739 KDPG and KHG aldolase, an Entner-Doudoroff

pathway enzyme which catalyzes creation of pyruvate and glyceraldehyde-3-phosphate

from 2-keto-3-deoxy-6-phosphogluconate (KDPG) is up-regulated by ~3-fold in the

∆cheY1 mutant, but not at all in the ∆cheB1cheR1 mutant. Additionally, abra_6420 PEP

carboxylase, which converts phosphoenolpyruvate to oxaloacetic acid, is up-regulated by

3-fold in the ∆cheY1 mutant only while abra_1950 glyceraldehyde 3-phosphate

dehydrogenase, involved in the glycolytic pathway, is down-regulated by 3-fold.

Carbon storage in the form of PHB granules may also be affected in both mutants.

One phasin family protein, abra_3170, is among the most abundant proteins in all

161

proteomes (attached spreadsheet Mutant_Proteome_Data), although it is of higher

abundance in both mutant proteomes. Examination of the up-regulated proteins in

∆cheB1cheR1 (Table 4.3) shows other phasin family proteins, abra_1168 and 1169, to be

up-regulated by 2.6-fold and 2.9-fold respectively, while in ∆cheY1 cultures abra_3170 is

upregulated by 3.2-fold, abra_1169 is upregulated by 7-fold and abra_1168 does not

show any fold-change (Table 4.3). Phasins are structural proteins embedded within the

lipid layer that surrounds poly-3-hydroxybutyrate (PHB) granules [186]. PHB granules

consist of a polyester core surrounded by a barrier layer thought to be composed of a lipid

monolayer with embedded and associated proteins that function to synthesize, store, and

access the PHB carbon when needed [186]. Azospirillum species are known to produce

small PHB granules under optimal conditions, with the amount of PHB increasing up to

30-fold when grown under conditions of high C:N ratio and high amounts of oxygen

[154]. During times of nutrient limitation, specifically when carbon sources are abundant

while nitrogen is limited, a bacterium can accumulate up to 80% of its dry weight in PHB

granules [186, 187]. It is thought that the presence of phasin proteins prevents individual

PHB granules from fusing, with phasin synthesis and abundance being highly correlated

to PHB synthesis and abundance [186]. Microscopic examination of cells shows the

presence of PHB granules in all cultures, both mutants and Sp7 control. However, the

amount of PHB in each was not quantified.

PHB synthesis is accomplished in a very simple pathway consisting of three steps

(Figure 4.2). Condensation of acetyl-CoA subunits to acetoacetyl-CoA is catalyzed by

beta-ketothiolase, while acetoacetyl-CoA-reductase catalyzes the subsequent formation of

162

Figure 4. 2 Pathway of PHB biosynthesis

Acetyl-CoA subunits are condensed to form acetoacetyl-CoA (catalyzed by beta-ketothiolase).

Acetoacetyl-CoA-reductase catalyzes the subsequent formation of hydroxybutyryl-CoA, and PHA

synthase then acts to polymerize individual units of hydroxybutyryl-CoA to form poly-hydroxybutyrate.

As discussed in the text, mutants lacking a functional CheY1 show upregulation of a number of proteins

annotated as acetoacetyl CoA reductase, although no proteins from this pathway were detected in any

cultures. In addition a number of phasin proteins, which surround PHB granules, are upregulated in this

mutant, suggesting an altered carbon metabolism.

Acetyl-CoA Acetyl-CoA

Acetoacetyl-CoA

Hydroxybutyryl-CoA

Polyhydroxybutyrate

Beta-ketothiolase (Condensation)

Acetoacetyl-CoA reductase (Reduction)

PHA synthase (Polymerization)

163

hydroxybutyryl-CoA. PHA synthase then acts to polymerize individual units of

hydroxybutyryl-CoA to form poly-hydroxybutyrate [186]. Gene sequences for those

Sp7 genes involved in PHB synthesis are present in the NCBI database. BLASTx [134]

comparison of these translated coding sequences to the Sp245 genome coding sequences

showed beta-ketothiolase (phaA) to correspond to abra_4231 with 100% similarity,

acetoacetyl-CoA reductase (phbB) to abra_4232 with 97% similarity, and poly-beta-

hydroxybutyrate synthase (phbC) to abra_4230 with 98% similarity. An additional gene

involved in PHB manipulation is PHB depolymerase (phaZ), corresponding to abra_0608

with 97% similarity. None of these genes were detected in any culture, either control or

mutant. However, 3 different proteins annotated as acetoacetyl-CoA reductase

(abra_5859, 5432, and 6359) were detected to be up-regulated by 3.0-, 2.9-, and 3.9-fold,

respectively, in ∆cheY1 mutants (Table 4.3).

It is perhaps expected for the ∆cheB1cheR1 mutant to have a greater amount of

PHB granules because of the larger size, especially if carbon storage in the form of PHB

is relative to the size of the cell, but the increase in phasin family proteins for the ∆cheY1

mutant is somewhat unexpected. Since carbon storage in the form of PHB granules is

known to increase in a high C:N ratio medium, which was not the case for these mutants,

the differential carbon storage in the form of PHB granules provides another clue that

these mutants may be sensing the environment differently than the wild type control.

Nitrogen and metal metabolism

Nitrogen regulatory protein P-II (abra_1565) is up-regulated by 5-fold in ∆cheY1

mutants (Table 4.3), just as it is in nitrogen fixing Sp7 cells, suggesting an increased need

164

for nitrogen sensing in these mutants. PhoH family protein (abra_0407) is upregulated in

both mutant cultures by 3-fold (Table 4.3). In enteric bacteria, PhoH is a phosphate

starvation inducible protein that possesses nucleoside triphosphate hydrolase activity. E.

coli PhoH is proposed to be part of the Pho regulon, whose expression is controlled by

two component system PhoR-PhoB [188]. The response to phosphate starvation and the

role of PhoH in A. brasilense or any other Azospirillum species has not been investigated.

However, the up-regulation of this protein in both mutants could suggest a role for this

PhoH family protein related to growth or clumping behavior in Azospirillum cells.

Cofactor biosynthesis

Both mutant cultures show a down-regulation of molybdopterin oxidoreductase

(abra_1181), although ∆cheB1cheR1 shows a far greater down-regulation (17-fold) than

does ∆cheY1 (3-fold, Table 4.4). Molybdopterin oxidoreductase complexes molybdenum

to pterin, thus activating it for use in more than 50 molybdate-containing enzymes [189].

Since the molybdenum cofactor is used in a diverse set of oxidation and reduction

reactions involved in nitrogen, sulfur and carbon metabolism [190], the down-regulation

of this one enzyme could potentially affect the function of many additional enzymes.

∆cheB1cheR1 mutants show a 5-fold down-regulation of cobalamin biosynthesis protein

CobW, one enzyme within the pathway responsible for synthesis of co-factor cobalamin

(vitamin B-12). Cobalamin is a co-factor for a number of other enzymes [191], and its

down-regulation suggests a related effect on the function of those enzymes dependent on

it for their function, which also suggests differences in the metabolic activity of

∆cheB1cheR1 versus ∆cheY1 and Sp7.

165

Respiratory chain

Differences in the requirement for (or expression of) respiratory chain

components is implied in the up-regulated proteins for ∆cheY1 and ∆cheB1cheR1 (Table

4.3). In ∆cheY1 mutants, abra_0437, an FeS binding protein, is up-regulated, just as it is

in nitrogen fixing Sp7 cells. Additionally, cytochrome c oxidase (abra_2759) and

cytochrome c’ (abra_2759) is up-regulated in ∆cheY1 cells. In contrast, ∆cheB1cheR1

mutants show up-regulation of abra_4337 cytochrome b/b6 and abra_4829 ubiquinol

oxidase, while ferredoxin abra_1305 is down-regulated, suggesting the use of a different

or additional respiratory pathway in ∆cheB1cheR1 mutants. Taken together, these data

suggest that each mutant is utilizing a different respiratory metabolism with the notable

similarity of ∆cheY1 to Sp7 nitrogen fixation. This is interesting as nitrogen fixing Sp7

cells are also usually shorter and rounder relative to cells grown under nitrogen-replete

conditions.

Signaling proteins

Interestingly, a number of signaling proteins are differentially expressed in the

two mutant cultures. ∆cheY1 shows up-regulation of chemotaxis sensory transducers

abra_3025 and abra_5341 while abra_2635 is down-regulated. One chemotaxis sensory

transducer, abra_2567, is up-regulated in both mutant cultures. Response regulator

receivers abra_0805, and abra_2692 are up-regulated in ∆cheY1, but no response

regulator receivers are found to be up- or down-regulated in ∆cheB1cheR1. PAS sensor

proteins abra_5331 and abra_3691 are up-regulated in ∆cheY1 mutants, while PAS

sensor protein abra_5409 is up-regulated in ∆cheB1cheR1. PAS domains are found in

cytosolic signaling transduction proteins that function to sense the intracellular energy

166

status of the cell [192]. They bind small co-factors such as heme or flavin or adenine,

with the small molecule bound allowing for specificity in sensory input [192, 193]. In

contrast to ∆cheY1 mutants, ∆cheB1cheR1 mutants show a greater than 5-fold up-

regulation of only one signaling protein, abra_6288. A BLASTp [52] search of the

abra_6288 translated sequence against the NCBI non-redundant protein database revealed

that it has 46% identity to a chemotaxis sensory transducer from Rhodopseudomonas

palustris BisB5, and contains a C-terminal methyl-accepting chemotaxis-like domain

with a HAMP region directly upstream of that domain, which indicates that this is a

chemoreceptor or MCP (Figure 4.3). HAMP (histidine kinase, adenylyl cyclase and

methyl carrier protein) regions are common signaling domains occurring in a wide

variety of signal transduction proteins in bacteria, and function both as extracellular

sensors and as dimerization domains [193]. The up-regulation of only this MCP

signaling protein in ∆cheB1cheR1 mutants suggests that this MCP has a special role that

is relevant or specific to the physiology of cells such as ∆cheB1cheR1. Notably, these

differences are also consistent with changes in the physiology of these cells that require

different sensing and signaling systems.

Proteins related to changes in cell morphology/clumping

As discussed earlier, ∆cheB1cheR1 mutant cells lack the methylesterase and

methyltransferase that functions in the adaptation pathway of the Che1 operon. Although

these mutants grow well and at a rate indistinguishable from that of the wild type cells,

they also produce EPS of different properties relative to the wild type cells [117] which

may correspond to less EPS or different types of EPS. This change in EPS production

167

Figure 4. 3 BLASTp depiction of conserved domains in abra_6288

The protein sequence indicates that this protein contains a C-terminal methyl-accepting chemotaxis-like domain with a Histidine kinase, Adenylyl cyclase,

Methyl-accepting protein, and Phosphatase (HAMP) domain. The presence of these two domains suggests that this protein may be involved in sensing

environmental changes and signaling signaling those changes. 5-fold upregulation of this signaling protein in only ∆cheB1cheR1 cells suggests that it may

play an important role in cells exhibiting a phenotype such as these.

168

also leads to the observed defect of ∆cheB1cheR1 mutant in clumping by cell-to-cell

aggregation, an event that is a prerequisite to flocculation, which is not observed in

cultures of the ∆cheB1cheR1 mutant. Further, they grow to longer length prior to cell

division, and divide at longer lengths than wild-type [117]. In contrast, ∆cheY1 mutant

cells lack the signaling output of the pathway, since CheY1 is homologous to the

response regulator CheY from E.coli that interacts with the flagellar motor to modulate

the direction of rotation of the flagellum. Whether CheY1 interacts with Azospirillum’s

flagellar motor is unknown but mutations in CheY1 affect the swimming motility bias of

cells [117] suggesting that it does interact with the flagellar motor. In addition, mutations

in CheY1 increase the propensity of cells for clumping, perhaps by modulating EPS

production (amount and/or composition). Individual A. brasilense ∆cheY1 mutant cells

also divide at shorter cell length relative to the wild type [117], as discussed earlier. All

the phenotypes displayed by the ∆cheY1 mutant strain are similar to that of a strain

lacking the entire Che1 pathway, suggesting that CheY1 functions as the signaling output

of the pathway to modulate clumping, cell length at division and chemotaxis, either

through direct interaction or through an indirect effect on other pathways.

With that in mind, one goal of this proteomics experiment was to identify changes

that could be common to both mutants, thus pointing to a single output of the Che1

pathway or unique to each mutant and thus pointing to unique proteins involved in

clumping or in increasing cell length. For instance, if a protein is found to be up-

regulated in ∆cheY1 mutant strain while remaining unchanged, down-regulated, or not

expressed in the ∆cheB1cheR1 mutant, then the possibility exists that it is related to

169

clumping behavior. Likewise, if a protein is found to be up-regulated in the ∆cheB1cheR1

mutant strain while at the same time remaining unchanged, unexpressed or down-

regulated in the ∆cheY1 mutant strain, the possibility exists that it is related to cell growth

or elongation. Those that are unique to each mutant whether up- or down-regulated are

probably related to unique cellular responses that result from mutation effect on cell

length and clumping and so are likely specific to the physiology of the mutant.

Cell wall biosynthesis

Phenotypic characterization of ∆cheY1 mutant cells indicate a shorter cell with an

increased tendency to flocculate and an altered colony morphology from wild-type,

suggesting differences in the outer cell wall composition and attached exopolysaccharide

(EPS) [117]. Up-regulation of a number of proteins in the ∆cheY1 mutant proteome

supports the hypothesis that ∆cheY1 mutants are undergoing extensive morphological

remodeling. Altered peptidoglycan synthesis in ∆cheY1 mutants is suggested by up-

regulation of abra_1829 lytic glycosylase while abra_3497 Serine-type D-alanyl-D-

alanine carboxypeptidase is down-regulated. Three-fold up-regulation of two

glycosyltransferases, one with phosphorylase activity (abra_3118), and another without

(abra_1786) suggests that either new lipid A or EPS components are being synthesized or

different sugars are being added in ∆cheY1 cells. Outer polysaccharide layer changes are

further suggested in ∆cheY1 mutants by the 2.6-fold up-regulation of polysaccharide

biosynthesis protein CapD (abra_1966), and 3.5-fold upregulation of abra_3118

glycogen/starch/alpha-glucan phosphorylase.

170

The cell envelope of gram-negative bacteria is composed of both an inner

membrane and an outer membrane separated by a periplasmic space containing

peptidoglycan [194]. Peptidoglycan is composed of long chains of alternating N-

acetylglucosamine and N-acetylmuramic acid subunits. These long glycan chains are

cross-linked together with short peptide bridges, forming a strong and flexible protective

structure surrounding the inner membrane. Lytic transglycosylases, such as abra_1829

which is up-regulated in ∆cheY1 mutants (Table 4.3), are enzymes that cleave

peptidoglycan in order to allow biosynthesis and turnover of peptidoglycan layers [195],

while D-Ala-D-Ala carboxypeptidases such as abra_3497, which is up-regulated in

∆cheY1 mutants, facilitate formation of peptide cross-links. Differential regulation of

these proteins in in ∆cheY1 mutants suggests these cells may be remodeling

peptidoglycan layer.

The inner membrane of the cell envelope is a phospholipid bilayer with embedded

transmembrane proteins and associated lipoproteins, while the outer membrane is an

asymmetric bilayer with only the inner leaflet composed of phospholipid [174]. The

outer leaflet of the outer membrane is a selectively permeable membrane composed of

lipopolysaccharides (LPS) with associated lipoproteins and imbedded proteins that act as

transport channels for small molecules less than 600 Da [183]. LPS is composed of

three elements: a complex glycolipid called lipid A which anchors the LPS to the

outermembrane, a core oligosaccharide that provides an attachment site for O-antigen,

and O-antigens which are composed of repeating units of oligosaccharides [174, 196].

Each component of LPS is synthesized individually via a pathway beginning with

activation of cytosolic sugars through the addition of UDP. Lipid A is synthesized in the

171

cytosol via the action of UDP-N-acetylglucosamine acyltransferase, deactylase and N-

acyltransferase. The synthesis pathway for the LPS core oligosaccharide inner region is

unknown, although the outer core region of hexose sugars is added to the inner core by

glycosyl-transferases which are specific for the sugar molecule added. ∆cheY1 mutants

show 3-fold up-regulation of two glycosyltransferases, abra_3118 and abra_1786, while

∆cheB1cheR1 do not show up-regulation of any glycosyltransferases, suggesting that

either new lipid A or EPS components are being synthesized or different sugars are being

added in ∆cheY1 cells. This observation is consistent with the phenotypic observation of

morphology differences between the two mutants.

Exopolysaccharides (EPS) are polymers of a wide variety of high molecular

weight polysaccharides and/or proteins that are loosely attached to the outer cell

membrane [197]. Capsular polysaccharides (CPS) appear as fibrous material at the cell

surface, anchored to the outer membrane via covalent linkages to phospholipid or to lipid

A. Synthesis and transport of EPS and CPS utilizes many of the same pathways and

proteins as those involved in LPS synthesis [197]. EPS is constructed from intracellular

nucleoside diphosphate sugar precursors. Some are synthesized through undecaprenol

intermediates, attached to a fatty acyl carrier and polymerized in the same manner as LPS

[183]. Others are not associated with lipids, but are instead polymerized by synthetases

in the cytosol and transported to the outer membrane via ABC-type transporters.

Azospirillum species synthesize both exopolysaccharides (EPS) that are loosely attached

and capsular polysaccharides (CPS) that are anchored to the outer leaflet of the outer

membrane [88]. Differences in synthesis or export of EPS and CPS in ∆cheY1 mutants is

suggested by the up-regulation of polysaccharide biosynthesis protein CapD (abra_1966)

172

and glycogen/starch/alpha-glucan phosphorylase (abra_3118) in ∆cheY1 mutants. CapD

is a gamma- glutamyltranspeptidase in gram positive Bacillus anthracis cells and

functions in anchoring polysaccharide layers to the peptidoglycan [198, 199]. In gram

negative Rickettsia prowazekii cells, CapD catalyzes the epimerization of UDP-glucose

to UDP-galactose, and is important in polysaccharide biosynthesis [200]. Glycogen

phosphorylase is involved in the degradation of glycogen, although the role of glycogen

storage in Azospirillum is still not well understood. Recently, an A. brasilense Sp7

mutant was created that lacked one glycogen phosphorylase gene termed glgP that had

97% similarity at a protein sequence level to abra_3118. Although the morphology of

this mutant was similar to wild-type, the mutant showed impaired biofilm formation,

decreased amounts and different composition of EPS, but greater capability for survival

under stress conditions such as starvation and osmotic pressure [201]. Up-regulation of

abra_3118 glycogen phosphorylase protein in ∆cheY1 mutants suggests that this protein

may play a role in ∆cheY1 mutant cells, either in using glycogen stores as an additional

carbon source, or in modifying EPS. Taken together, the above data suggests that

∆cheY1 mutants are engaged in active remodeling of their cell wall and attached EPS,

perhaps synthesizing more EPS and LPS than wild-type cells.

Under conditions of nutrient limitation when carbon sources are abundant (high

C:N ratio), Azospirillum species will aggregate, with accompanying loss of motility, and

formation of PHB granules and a thick polysaccharide-rich coat or capsule [202]. Extent

of flocculation and/or aggregation is dependent upon the composition of the EPS

surrounding the cells [87, 154]. Although both ∆cheB1cheR1 and ∆cheY1 cultures were

grown under high aeration conditions in minimal media containing both carbon and

173

nitrogen sources (not conditions to induce flocculation and aggregation), ∆cheY1 still

exhibits an increased tendency to clump, while ∆cheB1cheR1 cells show the opposite

tendency and are in fact impaired in flocculation. The above mentioned results, taken

together with up-regulation in ∆cheY1 mutants and no change in ∆cheB1cheR1 mutants,

suggest a number of possible proteins that could contribute to a morphological change

leading to clumping in ∆cheY1 mutant cells.

Type VI secretion systems

An interesting set of proteins shown to be up-regulated in the ∆cheY1 mutant but

not detected in the ∆cheB1cheR1 mutant includes components of the Type VI secretion

system (T6SS), expressed as shown in Table 4.5. Type VI secretion systems were only

recently recognized and named as a separate secretion system for export of proteins either

with or without leader peptides into the extracellular media [203]. The primary

characterization of components of this system has been done in pathogenic bacteria due

to the requirement of the T6SS for virulence in pathogenic strains of E. coli, Vibrio

cholerae, Vibrio anguillarum, Salmonella typhimurium and Pseudomonas aeruginosa, to

name a few [204]. Nevertheless, only a few of the proteins found within large genomic

regions encoding the T6SS components have been characterized, and identity of proteins

secreted by this system remains elusive [204].

The Azospirillum brasilense Sp245 genome shows 2 regions containing genes

annotated as Type VI secretion components. One small region contains genes encoding

IcmF (Intracellular multiplication F, abra_0646) and a DotU family protein (Defect in

organelle trafficking, abra_0647), both involved in Type 4 secretion systems as well,

174

arranged as shown in Figure 4.4. Another large genomic region contains conserved T6SS

components (discussed below) Vgr family protein (Valine-lysine repeat protein,

abra_4434), ClpV1 Atp-ase (abra_4435), EvpB (abra_4440) and ImpA (abra_4455),

arranged as depicted in Figure 4.5. No Hcp (Hemolysin coregulated protein) protein, a

conserved secreted T6SS component, is found within the Sp245 genome structure,

although a number of proteins annotated as “hemolysin” are present within the genome.

In both Sp7 and in ∆cheY1 mutants, seven known components of the latter T6SS are

expressed, including those components known to be necessary for T6SS function (ImpA,

ClpV1, and EvpB) with abra_4441 VCA0107 and abra_4455 ImpA being up-regulated in

∆cheY1 mutants over Sp7 control cells (Table 4.5). Additional hypothetical proteins and

a protein of unknown function DUF796 are also detected in ∆cheY1 cultures. In contrast,

only abra_4428 σ54

interaction domain containing protein and abra_4452 Type VI

secretion protein VCA0014 family are detected in ∆cheB1cheR1 cultures, while no other

components are detected (Table 4.5). In Sp7 control cultures as well as in each mutant

culture, one protein from the former region containing genes for IcmF and DotU are

detected. Nearby hypothetical protein abra_0639 and dctP TRAP dicarboxylate

transporter abra_0643 are also detected in all cultures, suggesting this operon is

expressed at similar levels in all cultures.

Type VI secretion systems were first discovered in Rhizobium leguminosarium in

1996 in mutants that were impaired in nodulation to pea plants [205] but not recognized

as a distinct and conserved secretion system until 2006 [206]. This protein excretion

system has now been discovered to be conserved among a wide range of bacteria and is

present in about 100 bacterial genomes, primarily in pathogenic bacteria and non-

175

Figure 4. 4 Schematic depiction of genomic region containing type VI secretion system homologs to IcmF (abra_0646) and and DotU (abra_0647)

Abra0645 is a type VI secretion associated protein of the BMA_A0400 family, while abra_0646 is an IcmF homolog, and abra_0647 is a DotU family

homolog. Abra_0646 IcmF homolog is upregulated in CheY1 mutants, but not in Sp7 or in CheB1CheR1 mutants. IcmF homologs contain a

transmembrane domain and a Walker A nucleotide binding motif, while DotU-like proteins contain a transmembrane segment in the C-terminal end, often

with C-terminal extensions that show similarity to OmpA proteins. T6SS secretory apparatus are thought to function through association of IcmF and DotU

family homologs.

OmpA domain T6SS associated DotU family IcmF

Abra_0645 Abra_0646 Abra_0647 Abra_0644

176

Figure 4. 5 Schematic representation of extended genomic region surrounding the type VI secretion system components in the Sp245 genome

Black line represents genomic DNA, while arrows are representative of the direction/strand on which individual genes are located. Each individual line

represents a possible grouping of genes in an operon structure based on distance between the genes. Dark aqua represents those components upregulated and

CheY, while dark blue indicates those gene products that were detected in Sp7 only. The number of sporulation domain proteins and hypothetical proteins in

this region combined with the upregulation of proteins only in smaller CheY cells present intriguing possibilities for the secretion system playing a role in

morphological changes related to clumping and cell size.

DUF796 Family

abra4439

T6SS lysozyme

abra4438

VC_A0110

abra4437

VC_A0111

abra4436

ClpV1 ATPase

abra4435

hypothetical

abra4433

Vgr Family

abra4434

Evp/VC_A0108

abra4440

hypothetical

abra4442

hypothetical

abra4443

VC_A0107

abra4441

T6SS VCA0114

abra4452

hypothetical

abra4451

Sporulation domain

abra4449

Lytic Transglycosylase

abra4448

Stage II SpoE

abra4447

HIS kinase-HAMP

abra4446

hypothetical

abra4445

hypothetical

abra4444

Sporulation domain

abra4450

ImpA

abra4455

DUF 796 Family

abra4456

hypothetical

abra4457

hypothetical

abra4454

hypothetical

abra4453

hypothetical

abra4458

A

A

C

C D

D

B

B

177

pathogenic soil bacteria of the proteobacteria subclass [207]. Found in loci of 12 – 30

genes, with components located both within operon structures and also in close proximity

to those operons, the T6SS is thought to mediate interaction with eukaryotic hosts [203]

and to be involved in processes such as adherence, cytotoxicity and survival and

persistence within a host cell [207].

Commonly recognized genes within T6SS encode proteins similar to those of

Legionella pneumophila Type IV secretion proteins, IcmF (intracellular multiplication F)

and DotU (defect in organelle trafficking) [203]. IcmF-like proteins, often called ImpA

like the IcmF homolog originally found in Rhizobium, contain a transmembrane domain

and a Walker A nucleotide binding motif, while DotU-like proteins contain a

transmembrane segment in the C-terminal end, often with C-terminal extensions that

show similarity to OmpA proteins [203]. Secretion function is dependent upon

association of IcmF and DotU homologs in most known systems, although precisely how

these two proteins interact and function to mediate secretion is not known [203]. As

stated above, the Sp245 genomic region encompassing abra_0644 to abra_0647 contains

genes encoding DotU and IcmF homologs, and all cultures show expression of genes in

this region (Table 4.5). An additional conserved gene found within T6SS loci encodes a

ClpB-family protein, ClpV, an ATP-ase whose function is unknown, although it is

thought to unfold proteins for transport through the secretory apparatus and to provide

energy for protein translocation [204]. Also common within T6SS loci are genes

encoding putative outer membrane lipoproteins, forkhead associated (FHA) domain

proteins, membrane-associated proteins and ATP-ases that may be contributing to

regulation and/or secretion function [208]. A second extended region within the Sp245

178

genome encompassing genes abra_4433 to abra_4458 contains ClpV ATP-ase (Figure

4.5), which is detected only in Sp7 and ∆cheY1 cells, but not in ∆cheB1cheR1 cells.

The only known proteins to be secreted from the T6SS are Hcp (Hemolysin co-

regulated protein), and VgrG (valine-lysine repeat protein G), which together are thought

to compose part of the secretion apparatus of T6SS systems. Hcp forms hexameric rings

that polymerize in vitro to form long tubes, and thus it is thought to function as a channel

for protein export to extracellular medium [204]. VgrG proteins have a conserved region

with similarity to T4 bacteriophage tail spike proteins, gp27 and gp5, which associate as

heterotrimers to form a needle or spike that facilitates injection of DNA into a target

bacterium [208, 209]. An additional conserved gene within T6SS clusters, VCA0109,

shows similarity to T4 bacteriophage gp25, a component of the baseplate for the

bacteriophage tail spike that serves to anchor it into the cell wall [208], making it likely

that this protein also participates in the structural secretory apparatus. Additional effector

proteins, RsbB ribose-binding protein in Rhizobium [203], and EvpB virulence factor

protein in Edwardsiella tarda [206], are also thought to be secreted from T6SS. Of these,

Vgr family protein (abra_4434), and EvpB (abra_4440) are found within the T6SS

extended region in the Sp245 genome, while Hcp protein is not found within the genome

at all. EvpB is expressed in both Sp7 and ∆cheY1 cultures, but is not detected in

∆cheB1cheR1 cultures (Table 4.5).

In Vibrio anguillarum, a pathogenic marine bacteria which causes hemorrhagic

septicemia in fish, the T6SS operon contains 8 genes, 4 that encode known components

of the T6SS and 4 that encode solute binding proteins, transport proteins and a D-ala-D-

179

ala ligase [210]. The four atypical T6SS proteins in this species function as either

positive or negative regulators of secretory function and control expression of

extracellular proteases via affecting the expression levels of RpoS involved in stress

response, and thus of VanT involved in quorum sensing [210], suggesting that it is

possible that proteins encoded by genes in close proximity to or within the T6SS operon

can serve a widely variant regulatory function. In the larger T6SS genomic region of

Azospirillum brasilense Sp245 a number of hypothetical proteins, proteins of unknown

function containing DUF domains, sporulation proteins, lytic transglycosylase and a His-

kinase HAMP domain protein are present (Figure 4.5). Proteins such as sporulation

domain proteins, lytic transglycosylase and HAMP domain proteins can be involved in

cell remodeling on a number of levels, and their position in close proximity of the genes

encoding theT6SS system suggest the intriguing possibility that this secretion system

may be involved, either directly or indirectly, in morphological change of the cell.

In other bacterial species, expression of T6SS genes are known to be tightly

controlled at a transcriptional level [203, 206, 207]. Transcription is regulated by

members of the AraC family of transcriptional regulators and by RpoN (σ54

) [207, 211],

and seems to be inversely related to transcriptional regulation of Type III secretion

system components [206, 207]. Interestingly, ∆cheY1 mutants show up-regulation of

abra_1953 spermidine/putrescine ABC transporter, a polyamine transport system (Table

4.3). Spermidine transport pathways have been shown to regulate expression of Type III

secretion systems in P. aeruginosa [212], and up-regulation of this transporter in ∆cheY1

mutants but not in ∆cheB1cheR1 mutants suggest it could play a role in the coordinated

regulation of the Type III and Type VI secretion systems. A fold-change level or even

180

detection of Type III secretion system components could not be determined because no

proteins in the database are annotated as such.

In microarray studies, expression of T6SS components in Vibrio cholerae have

been shown to be controlled through the action of the flagellar regulatory network, with

motility and virulence being oppositely regulated [211]. Expression of T6SS components

in vitro in Pseudomonas aeruginosa cultures can be induced by deleting the RetS gene,

part of a two component system important in establishing chronic infection in the host

cells [203]. Taken together, the above data suggests that T6SS expression can be

controlled by two-component systems, and may be related to a variety of cell morphology

changes such as those observed in the ∆cheY1 mutants. Further, the abundant presence

of T6SS components in ∆cheY1 mutant cells in comparison to ∆cheB1cheR1 mutant cells

suggests that the Che1 operon is functioning, either directly or indirectly, to modulate

expression of T6SS components, although further studies are needed to investigate this

possibility.

Conclusions

Proteomics investigations give a snapshot picture of the proteins present at a

given moment in time in the dynamic life of a cell. In this study we have examined the

proteomes of mutants in the Che1 chemotaxis pathway grown to early mid log phase, and

compared the detected proteins to those detected in a wild type Sp7 culture. Although all

cultures were grown under the same conditions, some significant differences emerge

between the detected proteome derived from the cheY1 mutant lacking a functional

forward signaling pathway, and that lacking an adaptation pathway (∆cheB1cheR1).

181

Earlier studies revealed significant physiological differences between the two mutants,

and the proteome further identifies differences in metabolic function between the two.

Patterns of up-regulation and down-regulation reveal very different metabolic

function, especially in carbon and amino acid metabolism. Of particular interest is the

increased abundance of phasin proteins, suggesting a tendency for the smaller ∆cheY1

mutant to store carbon, and perhaps indirectly suggesting an altered sensing of the

environment in ∆cheY1 cells. An Azospirillum Sp7 PHB synthase mutant strain was

constructed and characterized [213]. This mutant exhibited longer generation times and a

reduced ability to deal with stress conditions and also had a reduced chemotactic

response. Further, it produced greater amounts of exopolysaccharides (EPS) and

exhibited increased aggregation [213]. This same phenomenon is noticed in ∆cheY1

cells, which show an increased tendency to clump and altered EPS composition and/or

amount.

The detection of type VI secretion system components in both Sp7 and ∆cheY1

mutants, with up-regulation noted in ∆cheY1 cells, is very interesting. Further,

∆cheB1cheR1 cells show only a few components of the same system, suggesting that the

Che1 operon two component signaling system has an effect, either direct or indirect, on

the expression or operation of the Type VI secretion system. Further, four components of

the same system are detected in Sp7 controls grown to late mid log phase, but none are

detected in nitrogen fixing cells. Further investigation is needed to help clarify the role of

this secretion system in Azospirillum cells.

182

Chapter 5. Development of a set of gateway-compatible

destination vectors containing C-terminal tags

Introduction

In previous chapters of this work, proteomic surveys of Azospirillum cell cultures

were conducted. These surveys yield a snapshot of proteins present at a single point in

time, but give no information about the functionality of these proteins. Proteins rarely

work alone within a cell, but instead function in association with other proteins. Thus,

the putative function of an unknown protein can possibly be inferred from its known

interacting partners. A number of hypothetical proteins and proteins of unknown

function are detected in Azospirillum proteomes and are further seen to be up-regulated

under different growth conditions or in chemotaxis mutants; however, as discussed, the

function of most of these proteins remains unknown. Expression of proteins fused to an

affinity tag provides an easy way to purify a variety of proteins using a uniform

purification scheme [80]. A number of vector systems have been developed and

optimized for heterologous expression and purification of fusion proteins from bacterial

systems such as E coli [214-217] and B. subtilis [218], and for investigation of protein

complexes in model systems such as E. coli [83] or S. cerevisiae [76, 79, 219]. However,

although E. coli can be used for heterologous expression of proteins from other bacteria,

exploration of protein interacting partners needs to be done in the system of interest (in

vivo) to avoid false positives and facilitate interpretations. Organism-specific interacting

partners may be uniquely present in a system of interest but may not have any homolog in

E. coli. Ideally and to be widely applicable, vector sets with broad host range expression

capabilities which allow ease of cloning and also provide flexibility in the choice of tag

183

employed are needed for investigation of protein complexes in systems other than E. coli

[85]. In this chapter, a set of vectors has been developed and tested to facilitate

investigation of protein-protein interactions on a genome wide scale in a wide variety of

non-model bacteria, for which dedicated vectors have not been developed.

Coupling affinity tag purification with mass spectrometry has been shown to be a

viable method for purifying protein complexes on a large scale and subsequently

identifying interacting partners using the tool of mass spectrometry [73, 74, 77]. A

single affinity tag fused to a protein allows for a homogeneous purification method but

can lead to a high degree of background contamination, making it difficult to determine

legitimate interacting partners [80, 220]. Tandem affinity tags allow for a much

“cleaner” purification than single step purification, and can also facilitate elution under

native conditions when a protease cleavage site is included [220]. Further, affinity tags

can influence the level of expression and the degree of solubility of a protein of interest in

an individual protein-dependent manner [85]. Therefore, in considering a large-scale

investigation of proteins and their interacting partners, it is desirable to have a set of

affinity tags for flexibility and increased coverage of protein expression and investigation

of interacting partners. Fodor et al [221] have created a modular broad host range vector

system allowing for insertion of different promoters and containing either single or

tandem (StrepII-FLAG) affinity tags for expression of proteins and investigation of

protein complexes in native bacterial systems other than E. coli [221]. However, most of

the vectors created have only a single affinity tag, and expression in bacterial systems

other than E. coli relies on change of promoter through cloning into restriction sites.

Cloning of a gene of choice into this set of vectors is done through restriction digest of

184

genomic DNA or of PCR products and subsequent ligation into complementary

restriction sites. Ligation of restriction-digested DNA into complementary vector

restriction sites can be a time-consuming and ineffective process, so vectors containing a

ligation-free cassette for cloning a gene of choice would definitely be an improvement.

Earlier work at Oak Ridge National Laboratory (ORNL) resulted in creation of a

vector based on the pBBRMCS5 [222, 223] parental vector that could be used for high-

throughput analysis of protein complexes in a bacterial system. These plasmids will

stably replicate in a number of gram-negative bacteria [224], making them a versatile

vector for expression and interaction studies in a wide range of bacterial species. To

facilitate ease of cloning for high-throughput applications, a ligation-free cloning cassette

used in Invitrogen Gateway® system [225] was included, along with a gentamicin

resistance marker. In this work, the numbers of vectors in this newly designed tool box

have been increased by inclusion of a variety of C-terminal tags for tandem affinity

purifications under native elution conditions, allowing for complementary protein

expression to give the greatest coverage of proteins and their interacting partners in a

bacterial system of interest.

Earlier work in investigating protein complexes in E. coli [74, 76] and in S.

cerevisiae [16, 77-79] utilized the technique of integration of a tagged version of the

protein into the genome in order to reduce the likelihood of non-physiological

interactions as a result of over-expression. In order to address this issue, we also

developed a set of vectors based on a pJQ200KS [226] parental vector. The p15A origin

of replication of pJQ200KS makes this parental vector non-replicative (and hence a

185

“suicide” vector) in non-enterobacterial hosts [227]. A gentamicin resistance cassette is

also included in this backbone vector for positive selection of those colonies possessing

the plasmid containing the gene of interest in enterobacteria, or for positive selection of

single crossover events in non-enterobacteria. To this backbone, we added the gateway

cloning cassette with one of three different C-terminal tags. In this chapter, development

of the above outlined set of vectors is chronicled, and characterization of the application

of this tool set from immunoprecipitation experiments conducted with both integrating

and exogenous expression vectors in Rhodopseudomonas palustris, an alpha-

proteobacterium, are presented.

Materials and Methods

Bacterial strains and growth

Three strains of E. coli were used. Library Efficiency® DB3.1™ Competent

Cells (Invitrogen, Carlsbad, CA) were used for vector backbone maintenance and

propagation. Subcloning Efficiency™ E. coli DH5α™ Competent Cells (Invitrogen,

Carlsbad, CA) were used for maintenance and propagation of entry and destination

clones. E. coli S17-1 [228] chemically competent cells were used for conjugative

transfer of pJQ200SK-based plasmids into Rhodopseudomonas palustris. DB3.1 cells

were grown on Luria-Bertani (LB) media containing 10 µg/mL gentamicin and 100

µg/mL chloramphenicol for positive selection of vector backbone with correct tag inserts.

DH5-α cells were grown on LB with either 10 µg/mL gentamicin for selection of positive

transformants of destination clones, or with 50 µg/mL kanamycin for selection of positive

transformants of entry clones. Plasmids are maintained in DH5-α cells, with glycerol

stocks maintained at –80°C.

186

S17-1 chemically competent cells were prepared by standard protocols [229].

Briefly, cells were grown in LB broth overnight. Two ml of overnight culture was used

to inoculate a 250 ml culture in LB containing 20 mM magnesium sulfate, which was

then grown to an OD600nm of 0.5. Cells were pelleted, and suspended in 20 mL of TFB-1

(30 mM potassium acetate, 50 mM manganous chloride tetrahydrate, 100 mM potassium

chloride, 10 mM calcium chloride dihydrate, 15% w/v glycerol, pH 5.8). After a 20

minute incubation, cells were gently pelleted and resuspended in 1ml TFB-2 buffer (10

mM MOPS, 75 mM calcium chloride, 10 mM potassium chloride, 15% w/v glycerol).

Aliquots were frozen with liquid nitrogen and stored at -80°C.

Transformation of plasmids into all E coli strains was done following standard

protocols [229]. Briefly, chemically competent cells were incubated on ice for 1 hour

with plasmid DNA, then heat shocked at 42°C for 45 seconds. After being placed on ice

for 2 minutes, LB media was added and cells were allowed to recover through incubation

with shaking at 37°C for 1 hour. Transformations were then plated out and incubated

overnight at 37°C. When selecting for parent vectors with individual tags added, E. coli

DB3.1 cells were grown on LB medium containing both 10 µg/ml gentamicin and 30

µg/ml chloramphenicol. When selecting for positive transformants containing vectors

with genes of interest, all strains were grown on LB medium containing 10 µg/ml

gentamicin.

R. palustris strain CGA010 was grown under photoheterotrophic conditions [30],

in defined minimal media with 10 mM succinate as a carbon source, with 100 µg/mL

gentamicin for selection of positive transformants or without antibiotics. Cells were

187

grown anaerobically in the light at 25°C, with stirring for large scale cultures. Growth

for small scale expression studies was done in 25 mL culture tubes, while 1.5 L cultures

were grown for immunoprecipitation experiments.

Transformation of pJQ200KS-based plasmids into R. palustris strain CGA010

was accomplished through mating with E. coli strain S17-1 using standard protocols.

Briefly, R. palustris cells were grown as described without antibiotics. S17-1 cells

containing the pJQ-based plasmid with a gene of interest were grown overnight in LB

media with gentamicin. After centrifugation and washing, S17-1 cells were mixed with

R. palustris cells in a 1:5 ratio and the mixture spread on LB plates overnight. Selection

for R. palustris clones containing the gene of interest within the genome structure due to

a single crossover event was accomplished by anaerobic growth on minimal media plates

containing 100 µg/mL gentamicin. Transformation of pBBR-Dest42-based plasmids into

R. palustris strain CGA010 was accomplished through electroporation, as described

elsewhere [11, 81].

Construction of pBBR-based destination vectors

The parent backbone vector was based on pBBR1MCS [223] with gentamicin

resistance, which was modified to include a Gateway cloning cassette followed by

tandem C-terminal V5 and 6xHIS epitopes and named pBBR5-Dest42 [81]. The

pBBR5-Dest42 vector backbone was restriction digested with BstB1 restriction

endonuclease (New England Biolabs, Ipswich, MA), to linearize the vector and remove

the HIS-V5 epitope. The resulting digest was run out on an agarose gel, and the 6000 bp

188

band representing the vector backbone was then gel-purified using a QIAquick Gel

Extraction Kit (Qiagen, Valencia, CA) following manufacturer’s protocols.

Amino acid sequences for the components of the TAP tag (ProteinA-TEV

protease site-Calmodulin Binding Protein (CBP)) were obtained from Rigaut et al. [80]

and those for the SPA (3xFLAG-TEV protease site- CBP) tag were obtained from

Zeghouf et al [75]. For the STH (StrepII-TEV protease cleavage site-6xHIS) tag, the

strepII sequence was obtained from Schmidt et al. [230]. A 4-cysteine motif (CCPGCC)

was added to facilitate detection of recombinant protein expression [84] and an additional

TEV protease site [231] was included in the tag constructs for more efficient proteolytic

cleavage of the outer tag. The modified affinity tags are termed TAP2, SPA2, and STH.

Amino acid sequences and molecular weights for each tag are given in Table 5.1. Amino

acid sequence for each tag construct was back-translated using VectorNTI (Invitrogen,

Carlsbad,CA) and codons optimized for expression in R. palustris, which possesses a

high-GC content genome. Codon optimization tables can be obtained at

http://genome.ornl.gov/microbial/rpal/. DNA sequences for the TAP2 tag, the SPA2 tag

and the STH tag were synthesized by GenScript (Piscataway, NJ) and placed in pUC

vectors [232]. Insertion of the newly designed C-terminal tags into parent backbone

vectors is illustrated by the flow chart presented in Figure 5.1. Tag sequences were PCR

amplified from each vector using M13 universal primers (M13 forward 5’ CGC CAG

GGT TTT CCC AGT CAC GAC 3’; M13 reverse 5’ GAG CGG ATA ACA ATT TCA

CAC AGG 3’). PCR product was then restriction digested using Cla1 restriction

endonuclease (New England Biolabs, Ipswich, MA), creating single stranded DNA ends

with complementary sequence to those created in the parent pBBR5-Dest42 after BstB1

189

Table 5. 1 C-terminal tags and sequences included in pBBR-based and pJQ200KS-based vectors

Tag

Name

Tag Motif Amino Acid Sequence* length Molecular weight

(kDa)

STH 4Cys – 2x StrepII tag – 2x

TEV protease site – 6x HIS

CCPGCCASAWSHPQFEKSGWSHPQFEKGGTGS

ENLYFQGGRGGSENLYFQGEGTGSHHHHHH

239bp 6.7

SPA2 4Cys- Calmodulin Binding

Protein (CBP) -2xTEV-

3xFLAG

CCPGCCASKRRWKKNFIAVSAANRFFKKISSSG

ALDYDIPTTASENLYFQGGRGGSENLYFQGELD

YKDHDGDYKDHDIDYKDDD

314bp 9.6

TAP2 4Cys- CBP- 2xTEV -

2xProteinA

CCPGCCASKRRWKKNFIAVSAANRFKKISSSGA

LDYDIPTTASENLYFQGGRGGSENLYFQGELKT

AALAQHDEAVDNKFNKEQQNAFAEILHLPNLN

EEQRNAFIQSLKDDPSQSANLLAEAKKLNGAQ

APKGVDNKFNKEQQNAFYEILHLPNLNEEQRN

AFIQSLKDDPSQSANLLAEAKKLNGAQAPK

632bp 21.2

* Underlined portions indicate the amino acid sequence of the motifs separated by lines in the motif area

190

Figure 5. 1 Flow chart of vector construction steps

The parent vector pBBR-Dest42 with HIS-V5 tandem tag was restriction digested with BstB1, while the

PCR product amplifying the tag construct was digested with restriction endonuclease Cla1, leaving

complementary overhanging DNA sequence for easier ligation. Tag sequence was then ligated into gel-

purifed parent backbone vector, with resulting ligation mixture transformed into DB3.1 E. coli cells.

Positive clones were confirmed by diagnostic restriction digest and DNA sequencing.

pBBR-V5-His vector

BstB1

T4DNA ligase

Cla1

PCR amplify insert

Transform into DB3.1 E coli

BstB1

C G A T

T A

A T

T A G C

Insert-pBBR vector

pUC-Tag insert vector

Confirm insert by

DNA sequencing

HIS-V5 tag

+ Gel Purification

C G A A

T T

T T

A A G C

191

restriction digest. The resulting fragment was ligated into the BstB1 site of the gel-

purified pBBR5-Dest42 vector using T4 DNA ligase (Promega, Madison WI) following

manufacturer’s instructions with the following exception. In order to increase yield, the

ligation reaction was allowed to proceed overnight at 17°C before transformation into

DH5α cells. Ligated parent vectors were transformed into DH5α E. coli cells following

standard transformation protocols [229] , as described above. Five colonies of each type

of backbone vector clone were chosen for overnight growth in liquid media and

subsequent DNA plasmid mini-prep (Qiagen, Valencia, CA). Positive clones were

confirmed by diagnostic restriction digest and by DNA sequencing using universal M13

primers.

Construction of pJQ200KS-based destination vectors

The pJQ200KS (5370bp) vector backbone was linearized by digestion with Xba1

restriction endonuclease (New England Biolabs, Ipswich, MA). DNA sequence

including the Gateway®

cassette (attR sites, chloramphenicol resistance cassette, and

ccdB gene) with the added C-terminal tag (~2100bp) was PCR amplified using T7

Universal primer (5’ TAA TAC GAC TCA CTA TAG GG 3’) and TagRevSeq (5’ CGA

CCG GGT CGA ATT TGC 3’). The resulting PCR product was digested with Xba1

restriction endonuclease (New England Biolabs, Ipswich, MA), and then ligated into the

vector backbone using T4 DNA ligase (Promega, Madison WI) following manufacturer’s

instructions with the following exception. The ligation mixture was incubated overnight

at 17°C in order to ensure a higher number of correct ligations. Ligated pJQ200KS-tag

vectors were transformed into DB3.1 E. coli cells following standard transformation

protocols [229]. Transformants were plated on LB-agar plates containing both 10 µg/ml

192

gentamicin and 30 µg/ml chloramphenicol, and five individual colonies were chosen for

plasmid DNA mini-prep as above. Positive clones for the base parent vectors containing

C-terminal tags were confirmed by restriction digest and sequencing.

Protein expression

Entry clones for individual genes had been created earlier [81], and a small set of

genes within these entry clones was chosen for expression tests using the newly created

vectors. Recombination reactions were performed between the entry clones and the

newly created parent vectors using LR clonase enzyme mix (Invitrogen, Carlsbad CA)

following manufacturer’s directions. To increase the yield of recombinants, the LR

reaction was allowed to proceed overnight at 17°C. Resulting destination clones (a set of

5 genes tagged with each of the three individual tags in pBBR-DEST42 parent vectors)

were transformed into DH5α E. coli cells, and plated on LB media plates containing 10

µg/ml gentamicin. The presence of a wild-type DNA gyrase in these cells allows for

negative selection of nonrecombinant plasmids. Expression of the ccdB gene found in the

Gateway®

recombination cassette in the parent vector interferes with the function of wild-

type DNA gyrase, thus preventing replication in cells with wild-type DNA gyrase such as

DH5α E. coli (Invitrogen). Five resulting colonies containing destination clones were

chosen for further expression studies. 15 mL cultures were inoculated from these

colonies, and grown overnight at 37°C with shaking. Plasmid mini-preps were done

using QiaPrep spin mini-prep columns (Qiagen, Valencia CA) and DNA concentration

was determined using the NanoDrop spectrophotometer (Thermo Scientific, Wilmington,

DE). Additionally, glycerol stocks were made using 1 ml of the overnight cultures.

193

Destination clones created from the LR recombination reaction were transformed

into R. palustris CGA010 cells as described above, either by mating for pJQ200KS-based

clones or by electroporation for pBBR-Dest42-based clones. Transformants were

initially plated on PM-10 plates [30] containing 100 µg/ml gentamicin for positive

selection of clones containing plasmid and grown anaerobically with light. Individual

colonies were then used to inoculate 25 mL cultures of PM-10 media [30] containing 100

µg/ml gentamicin. Small cultures were harvested by centrifugation at 2000 xg for 5

minutes in a Sorvall tabletop centrifuge, washed, and divided into 2 equal aliquots. One

pellet was frozen at -80°C, while the other was lysed using 50 µl of PBS containing

Bugbuster® reagent (Novagen), 1 µl/ml benzonase nuclease (Novagen), lysozyme, 1

mg/ml leupeptin and 1% PMSF. An aliquot of lysate was run out on a Precise protein 10-

20 % gradient SDS poly-acrylamide gel (Pierce Biotechnology, a division of Thermo

Scientific, Rockford, IL) and used for characterization of expression. Western blot

analysis was performed using mouse anti-FLAG antibody (Sigma-Aldrich, St Louis MO)

for the primary incubation for SPA2 tagged proteins, rabbit anti-CBP antibody (Upstate,

a division of Millipore, Billerica MA ) for TAP2 tagged proteins, and mouse anti-HIS

antibody (Sigma-Aldrich, St Louis MO) for STH tagged proteins. Secondary incubation

was done either with goat anti-mouse or goat anti-rabbit IgG antibody complexed to

horseradish peroxidase as appropriate depending on the initial antibody incubation.

Fusion protein expression was visualized after DAB (3,3´-diaminobenzidine

tetrahydrochloride) reaction with horseradish peroxidase following manufacturers

recommendations (Pierce Biotechnology).

194

Once protein expression was confirmed for tagged proteins, colonies from fresh

PM-10 plates containing 100 µg/ml gentamicin were used to inoculate 25 mL R palustris

PM-10 starter cultures. Starter culture tubes contained 100 µg/mL gentamicin, while

large cultures did not contain antibiotics. Starter cultures were used to inoculate 1.5 L

production cultures, which were then grown to mid-log phase under anaerobic conditions

in the light with stirring as described above. Cultures were harvested at an OD660 of 0.5 –

0.7. Cells were pelleted by centrifugation at 6500 xg, then washed 3x with phosphate-

buffered saline (PBS). After resuspension in PBS, cells were split in two equal aliquots

and re-pelleted. Pellet from one aliquot was frozen at -80°C, while the other pellet was

resuspended in appropriate lysis buffer at a rate of 0.5 ml lysis buffer/ 0.1g cell pellet

weight. Lysis buffers contained 1x Bugbuster® reagent for chemical lysis of cells diluted

in binding buffers compatible with initial purification steps (M2 buffer for SPA2 and

TAP2 tagged proteins, BF3-FT buffer for STH-tagged proteins, Table 5.2). Benzonase

nuclease (1 µl/ml, Qiagen) was included in lysis buffers in order to digest genomic DNA

released on cell lysis. PMSF (10 mg/ml) and leupeptin (10 mg/ml) were included for

inhibiting protease activity, and lysozyme was included to help degrade the cell wall.

Additionally, beta-mercaptoethanol (0.2 µl/ml) was added to the lysis buffer in order to

break up disulfide bonds. Cell suspension was incubated on a rotating wheel for 1 hour

at room temperature. Lysate was cleared through centrifugation in a Sorvall centrifuge at

18000 xg for 20 mintues at 4°C, and supernatant transferred to a clean 15mL falcon tube.

Cleared lysates were frozen at -80°C for later immunoprecipitation experiments.

195

Table 5. 2 Buffer composition for each affinity purification step

Purification step SPA and TAP tag purification

buffer

STH tag purification buffers

Lysis Lysis Buffer: 10X Bugbuster reagent

M2 Binding Buffer

Benzonase Nuclease (25U/ul stock)

50uL

Lysozyme

100ug/ml PMSF

10 ug/ml Leupeptin

Lysis Buffer: 10X Bugbuster reagent

M2 Binding Buffer

25U Benzonase Nuclease

Lysozyme

100ug/ml PMSF

10 ug/ml Leupeptin

Initial affinity

purification step Flag or IgG binding -M2 buffer:

10 mM Tris-HCl @ pH 8.0

100 mM NaCl

10% glycerol

0.1% Triton-X 100

HIS binding buffer – BF3-FT

50 mM Tris-HCl @ pH 8.0

150 mM NaCl

50 mM NaH2PO4

10 mM imidazole

10% glycerol

Tev protease

cleavage TEV cleavage buffer

50 mM Tris-HCl @ pH 7.9

100 mM NaCl

0.2 mM EDTA

1mM DTT

0.1% Triton X-100

TEV cleavage buffer

50 mM Tris-HCl @ pH 7.9

100 mM NaCl

0.2 mM EDTA

1mM DTT

0.1% Triton X-100

Second affinity

purification step Calmodulin Binding Buffer: 10 mM Tris-HCl @ pH 7.9

100 mM NaCl

2 mM CaCl2

10 mM 2-mercaptoethanol

StrepII-tag Binding Buffer 50 mM Tris-HCl @ pH 7.9

100 mM NaCl

0.2 mM EDTA

1mM DTT

0.1% Triton X-100

Wash after 2nd

affinity

purification

Calmodulin Wash Buffer 10 mM Tris-HCl @ pH 7.9

100 mM NaCl

0.2 mM CaCl2

10 mM 2-mercaptoethanol

StrepII-tag wash Buffer 100 mM Tris-HCl @ pH 7.9

150 mM NaCl

1 mM EDTA

1 mM DTT

Elution Calmodulin Elution Buffer: 10 mM Tris-HCl @ pH 7.9

100 mM NH4HCO3

3 mM EGTA

10 mM 2-mercaptoethanol

StrepII-tag Elution buffer 100 mM Tris-HCl @ pH 8

150 mM NaCl

1 mM EDTA

20 mM desthiobiotin

196

Immunoprecipitation

SPA2 and TAP2 affinity purification was done according to protocols established

earlier [11]. STH-tagged proteins were affinity purified following protocols established

earlier [84], modified for use in bacterial cells and for batch processing mode. Aliquots

(200 µl) of resins for initial tag capture (Table 5.3) were washed three times in

appropriate binding buffer, and resuspended in appropriate binding buffer to create a 50%

bead slurry. Cleared lysates were incubated with bead slurry for 2-4 hours with 200 µl of

washed beads for affinity capture of proteins via the outer tags. For TAP-tagged proteins,

IgG Sepharose™ 6 Fast Flow resin (Amersham Biosciences) was used for Protein A

capture, while Anti-FLAG® M2 Affinity Gel beads (Sigma-Aldrich) were used for

3xFLAG capture in SPA tagged proteins. Nickel-NTA agarose resin (Qiagen) was used

for initial purification of STH-tagged proteins. Resins used for each step of affinity

capture for each tag are presented in Table 5.3. After incubation, resins with associated

proteins were washed 3 times with appropriate binding buffer, and a final wash with 1

mL of TEV protease buffer. After the final wash, the beads were centrifuged at 1500rpm

for 30 seconds, all of the wash buffer removed, and beads resuspended in 200 µl TEV

protease buffer. Ac-TEV protease (10 µl, 50U, Invitrogen, Carlsbad, CA) was added to

the bead suspension, and the suspension was then incubated on a rotating wheel at room

temperature for 30 minutes. Incubation was continued overnight on a rotating wheel at

4°C due to the dramatic reduction in efficiency of TEV protease cleavage at low

temperatures. After this incubation, proteins and associated binding partners are present

in the supernatant.

197

Table 5. 3 Resins used for capture in each affinity purification step

Affinity purification Initial (outer tag) capture Second (inner tag) capture

SPA affinity purification FLAG tag capture:

Anti-FLAG M2 agarose

beads (Sigma-Aldrich)

CBP tag capture:

Calmodulin Sepharose 4B

(Amersham Biosciences)

TAP affinity purification Protein A tag capture:

IgG sepharose beads

(Amersham Biosciences)

CBP tag capture:

Calmodulin Sepharose 4B

(Amersham Biosciences)

STH affinity purification 6xHIS tag capture:

NiNTA beads (Qiagen,

Valencia, CA)

StrepII tag capture:

Strep-tactin beads

(IBA, St Louis)

198

The second step of affinity purification began with washing a 200 µl aliquot (per

sample) of resins required for capture of the only remaining inner tag (CBP for both

TAP2 and SPA2 fusion proteins, StrepII for STH fusion proteins). After 4 washes with

appropriate binding buffer, bead slurries were centrifuged at 1500rpm for 30 seconds and

final wash buffer removed. For CBP capture, the beads were resuspended in 200 µl of

appropriate binding buffer, and 1.2 µl of CaCl2 (1M) was added. For StrepII capture,

beads were resuspended in StrepII binding buffer. Bead slurries from the TEV protease

cleavage step were centrifuged at 1500rpm for 30 seconds to pellet the beads.

Supernatant from the TEV protease cleavage step was added to the appropriate bead

slurry for secondary capture. Suspension was incubated at 4°C for 3 hours.

Elution was accomplished in three steps. The bead slurries were washed 4 times

in appropriate wash buffer. Elution for TAP2 and SPA2 tagged proteins included

chelation of calcium from the beads as calmodulin binding protein requires calcium for

binding to affinity resins. So, elution buffers included 3 mM EGTA, a chelating agent

that is specific for calcium. Elution for StrepII tagged proteins is based on replacing the

tagged protein bound to the streptavidin resin with biotin in the elution buffers. Although

the manufacturer’s protocol suggested 2 mM biotin for elution, we found that

concentration to be ineffective for elution. Therefore, we used elution buffers containing

20 mM biotin, which proved to be more effective for elution of tagged protein from the

resin. Elution buffer (100 µl) was added to washed beads with protein bound, and

incubated at room temperature for 10 minutes. Beads were then pelleted by

centrifugation at 1000 xg for 30 seconds, and supernatant was removed to a clean

199

centrifuge tube. This step was repeated twice more, and the three eluates were pooled for

a total of 300 µl, and frozen at -20°C until digestion.

Proteolytic digestion with sequencing-grade trypsin (Promega) was accomplished

through the addition of 1 µg of trypsin directly to the eluates. The pH was checked to

ensure the eluates were at a proper pH between 7 and 8. Calcium chloride (13 mM) was

added to the SPA2 and TAP2 eluates due to the presence of EGTA in the eluates.

Although trypsin will digest protein without calcium present, specificity is improved in

the presence of calcium. For this reason calcium was replenished in those samples.

Digestion was allowed to proceed overnight at 37°C. Proteolytic peptide samples were

dried down in a vacuum centrifuge and buffer exchanged into Buffer A (95% HPLC-

grade water, 5% acetonitrile, 0.1% formic acid). Samples were frozen at -80°C for later

analysis by mass spectrometry.

LC-MS/MS

Analysis of proteolytic peptides was accomplished via an automated one-

dimensional chromatographic separation followed by mass spectrometry of eluted

peptides. Chromatographic setup consisted of a FAMOS autosampler for injection of

individual samples to a C-18 reverse phase trapping column, followed by elution to a

resolving column via a Switchos and UltiMate HPLC (LC Packings, Sunnyvale, CA)

which was coupled directly to a nanoelectrospray source (Proxeon Biosystems, Odense,

Denmark) in line with an LTQ linear ion trap mass spectrometer (Thermo-Finnigan, San

Jose, CA). The resolving column consisted of 18 cm of 3 µm Jupiter C18 reverse phase

material (Phenomenex, Torrance, CA) packed via pressure cell into 100µ ID fused silica,

200

with a tip pulled to an opening of 5 µm (PicoTip, New Objective). This column was

directly coupled to the LTQ, and peptides were directly eluted from this column into the

mass spectrometer. Samples were run in triplicate in random order, with a single blank

sample of buffer A (95% water, 5% acetonitrile, 0.1% formic acid) run between each

immunoprecipitation sample in order to minimize carryover between samples. Peptides

were eluted from the resolving column in a gradient elution of 0-50% buffer B (30%

HPLC-grade water, 70% acetonitrile, 0.1% formic acid). The LTQ was operated under

control of Xcaliber software. Data was collected in data dependent mode, with 1 full

scan followed by 5 dependent scans in which 2 microscans were used to create each

spectrum. Dynamic exclusion was employed, with a repeat count of 1, exclusion list size

of 300 and a repeat duration of 180.

Data analysis

Experimentally derived mass spectra were searched against a database of amino

acid sequences from the CDS of R. palustris (located at http://compbio.ornl.gov/rpal_

proteome/databases) using SEQUEST search algorithm [3]. The database was created by

first appending a list of common contaminants and then reversing the amino acid

sequences and concatenating the reverse sequences to the forward sequences. One tryptic

cleavage site was specified for the searches. DTASelect [54] was used to filter and sort

the data, with default values (Xcorr of 1.8 for +1 peptides, 2.5 for +2 peptides, and 3.5 for

+3 peptides, and deltCN of 0.08) being employed. Tables were compiled in Microsoft

Access.

201

Results

A set of vectors containing three different C-terminal tandem affinity tags with

the backbone based on the pBBR-Dest42 parent vector [81] has been created. As shown

by the plasmid map in Figure 5.2, pBBR-Dest42 parent vector offers several advantages,

including a pBBR replicon for expression in a wide variety of bacterial species, a

Gateway® compatible cloning cassette (Invitrogen, Carlsbad, CA), gentamicin resistance

for positive selection of clones, and a mobilization region for conjugative transfer to other

bacterial species. C-terminal affinity tags included modified Tandem Affinity

Purification (TAP –2xCalmodulin Binding Protein(CBP) -Tobacco Etch Virus (TEV)

protease site –2xProtein A) tag [80], a modified Sequential Peptide Affinity (SPA-

2xCBP - TEV protease site - 3xFLAG epitope) tag [75] or a StrepII-TEV protease site-

6xHIS (STH) tag [233], as shown in Figure 5.3. A further advantage to the newly created

tag systems is the inclusion of a tetra-cysteine epitope that allows for easy detection of

protein expression in lysates, eliminating the necessity of performing a time-consuming

western blot procedure.

A second set of vectors was created based on the parent vector, pJQ-200KS [226],

which contains a p15A origin of replication, making it non-replicative in non-

enterobacteria. Thus it is a “suicide plasmid” which integrates into the genome of non-

enterobacteria, allowing for expression of tagged protein from a native promoter

sequence. The entire cloning cassette including the Gateway cloning cassette combined

with the C-terminal tandem affinity tags described above was placed in the pJQ-200KS

backbone vector. Tag sequences were confirmed for both pBBR-Dest42 based vectors

202

Figure 5. 2 Schematic drawing of the initial pBBR-Dest42 parent vector The parent vector backbone used for this work was derived from the pBBR1MCS5 vector. Modifications

to the initial pBBR1-MCS5 vector included addition of a gateway cloning cassette containing a

chloramphenical resistance gene and a ccdB gene flanked by lambda bacteriophage attR attachment sites.

A tandem affinity tag consisting of 6xHIS residues followed by a V5 epitope tag was added C-terminal to

the cloning cassette. The tandem affinity HIS-V5 tag was flanked by BstB1 restriction sites such that the

tag could be removed and a repertoire of different tags inserted.

pBBRDEST-42

6603 bp

REP

Mob

GM(R)

CM(R)

CcdB

V5 epitope

6xHis

T7 P

attR1

attR2

XbaI (3151)

BstBI (4901)

BstBI (5408)

203

Figure 5. 3 New set of pBBR-Dest42-based vectors showing expansions of the three new C-terminal tag constructs

pBBRDest42 parent vector contains all aspects of the pBBRMCS5 parent vector (genes involved in mobilization of the vector (mob) allowing for

conjugative transfer of the vector from a strain containing tra ????? genes in trans, gentamicin reisistance genes and rep genes involved in the replication of

the plasmid) in addition to the attR sites for Gateway compatible recombination flanking the cassettes for chloramphenical resistance and ccdB gene for

positive selection of transformants. The C-terminal 6xHis-V5 combinatorial tag was removed by restriction digest with BstB1, and the tags shown at left

were digested with CLa1 restricion endonuclease and subsequently ligated into the complementary BstB1 site of the digested parent vector. Correct

orientation of the tag into the parent vector was confirmed by restriction digest and by sequencing.

Order-4C-2Strep-6HIS

239 bp

4C6xHISTEV 1 TEV 2

StrepII 1StrepII 2ClaI (11) ClaI (228)

Order-4C-CBP-2TEV-FLAG

314 bp

4C

TEV 1 TEV 2 3xFLAGCBP

ClaI (11) ClaI (303)

Order-TAP-CBP-2TEV-2ProtA

632 bp

4C

CBP TEV 1

TEV 2

ProtA 1 ProtA 2

ClaI (11) ClaI (621)

pBBRDEST-42

6603 bp

lac operatorlac operator

REP

Mob

GM(R)

CM(R)

CcdB

V5 epitope

6xHis

T7 P

Plac

attR1

attR2

BstBI (4901)

BstBI (5408)

204

and for pJQ200KS-based vectors by both restriction digest and DNA sequencing as

described above.

Expression of RpoA, the 36.5 kDa alpha subunit of RNA polymerase, was

accomplished in all pBBR-Dest42 based vectors with all tags. Immunoprecipitation

experiments were also successful with all RpoA fusion proteins with all tags, with both

bait proteins and interacting partners detected via mass spectrometry (Table 5.4). The

higher molecular weight beta prime subunit (155.2 kDa) of RNA polymerase (RpoC) was

successfully expressed only in combination with the lower molecular weight tags, both

SPA2 and STH. Expression of an RpoC-TAP2 fusion from pBBR-Dest42 based

plasmids was unsuccessful. RpoC fusion proteins with lower molecular weight tags

were successfully purified and bait proteins with associated interacting partners were

detected via mass spectrometry. Sequence coverage of detected proteins and interacting

partners are presented in Table 5.5. Interestingly, although bait protein was detected in

all replicates of all tags, the SPA2 and TAP2 tags resulted in greater sequence coverage

for both bait proteins and interacting partners than did the STH tag. TAP and SPA tags

were originally tested and optimized for use in bacteria and/or yeast, while the STH tag

was optimized for use in mammalian systems.

Integrating vectors were tested only with an rpoA gene (rpa_3226) fused to a

SPA2 tag. Expression of fusion protein RpoA-SPA2 was confirmed via western blot and

mass spectrometry. Both bait protein and interacting partners were identified in mass

spectrometry experiments, although no discernible difference in the numbers of non-

specifically bound proteins was detected between the integrated version of RpoA-SPA2

205

Table 5. 4 Sequence coverage of interacting partners for RpoA (rpa3226) detected using mass spectrometry

3226 Interacting partners (http://string.embl.de/newstring_cgi/) RpoA 3226-

Gene Description -SPA -STH -TAP -pJQ-SPA

rpa3268 rpoB, DNA-Directed RNA polymerase subunit beta 21 2 29 1.5

rpa3267 rpoC, DNA-directed RNA polymerase subunit beta prime 10 2 18 1

rpa3227 rpsK, 30S ribosomal protein S11 ND ND ND ND

rpa3225 rplQ, 50S ribosomal protein L17 ND ND 14.4 ND

rpa3230 secY,preprotein translocase SecY ND ND ND ND

rpa1288 rpoD, RNA polymerase sigma factor ND ND ND ND

rpa3228 rpsM 30S ribosomal protein S13 ND 10.2 ND ND

206

Table 5. 5 Sequence coverage of interacting partners for RpoC (rpa3267) detected via mass spectrometry

3267 Interacting partners (http://string.embl.de/newstring_cgi/) RpoC 3267-

Gene Description -SPA -STH

rpa3268 rpoB, DNA-Directed RNA polymerase subunit beta 38 2

rpa3226 rpoA, DNA-directed RNA polymerase subunit alpha 45 6

rpa2693 relA, GTP pyrophosphokinase ND ND

rpa3269 rplL, 50S ribosomal protein L7/L12 ND ND

rpa2886 pyrG, CTP synthase ND ND

rpa1288 rpoD, RNA poymerase sigma factor 14.5 ND

rpa3270 rplJ, 50S ribosomal protein L10 ND 8.6

rpa3056 ndk, nucleoside diphosphate kinase ND ND

rpa0438 nusA, putative NusA protein transcriptional terminator ND 2.3

rpa2692 rpoZ, DNA-directed RNA polymerase subunit omega 23 ND

207

and the exogenously expressed RpoA-SPA2. Interestingly, much lower sequence

coverage was noted for RNA polymerase subunits when using an integrating vector than

when using a plasmid based strategy (Table 5.4), although lysis and immunoprecipitation

protocols were identical.

Single crossover events are the most common occurrence when using pJQ200KS

in non-enterobacteria [226]. As a result of a single crossover event, the pJQ-SPA2

plasmid would be integrated into the genome sequence in its entirety, with the tagged

version (including a stop codon) being translated while the untagged version is not. After

a single crossover event, PCR amplification of the genomic region flanked by upstream

genomic sequence and by common DNA sequence from the tag region should provide a

good diagnostic tool for detection of plasmid integration into the genome. Band size

obtained from this diagnostic PCR should reflect the size of the integrated gene plus the

size of a partial tag sequence. For instance, the expected size of the rpa3266 (rpoA) gene

(1019 base pairs) plus a partial SPA tag sequence including the 4C epitope plus the

Calmodulin Binding Protein epitope (261 base pairs) is 1280 base pairs. In order to

confirm integration of the tagged version of the gene into the genome, colony PCR of R.

palustris RpoA-SPA2 clones was attempted using a 5’ forward primer corresponding to

genomic sequence upstream of the rpoA gene and a 3’ reverse primer corresponding to

DNA sequence in the TEV protease site. A control consisting of colony PCR of E. coli

cells containing the pJQ200KS-SPA2 plasmid with an rpoA gene was performed in

tandem with the samples, and the results run out on an agarose gel. Unfortunately, only

control bands were present on the agarose gels, so establishment of integration of the

SPA2 tagged version of the rpoA gene into the genome was unsuccessful.

208

Lastly, a test of detection of protein expression was attempted using the 4C

epitope in combination with FlAsh reagent (Invitrogen). Lysates were combined with

FlAsh reagent and proteins were separated on an SDS-PAGE gel, following

manufacturer’s directions. After unsuccessful attempts to visualize protein presence

when using samples that had given positive results using a western blot protocol, a test

was done to determine whether the Bugbuster® reagent (Novagen) used for chemical

lysis of cells was interfering with the fluorescence of the dye. To test for interference of

lysis reagent with fluorescence, M2 lysis buffer containing either Bugbuster® or water

was used to dilute fluorescein dye. Molar concentrations of fluorescein varied from

picomolar (10 -12

) to millimolar (10-3

) concentration, representing concentrations of

protein possibly found within affinity purified samples. Dye solutions were spotted onto

glass microscope slides and visualized with a Bio-Rad fluorescent imager using Quantity

One software. Results indicated that fluorescence was dramatically reduced by the

Bugbuster® reagent, with molar amounts less than micromolar (10-6

) not being detected

in the Bugbuster® solution, while nanomolar (10-9

) concentrations were detected in the

water solution. Mechanical lysis methods (3 different sonication protocols and freeze-

thaw methods) in lysis buffers which did not contain detergent had been attempted earlier

and were unsuccessful, with little or no protein detected for any fusion protein. Due to a

lack of time, this methodology was not attempted again, so the detection of protein

expression in cell lysates via the 4C epitope remains untested for these tags.

Discussion

Proteins have a wide variety of sizes, sequences and structures. Thus, uniform

purification strategies such as those used in isolation of DNA are likely to be

209

unsuccessful in isolating a large number of proteins. Instead, the wide variety of protein

sequences and structures require optimization of purification strategies for each

individual protein. Addition of affinity tags to a protein allows for simplified generic

purification strategies that target the tag motif and can facilitate isolation of proteins and

their interacting partners on a genome-wide scale. The parental plasmid used in this

chapter, pBBR-Dest42 [81] offers a number of attractive advantages for protein

expression. First, pBBR-Dest42 (shown in Figure 5.2) is based on the pBBR1MCS5

plasmid, a medium copy number plasmid containing a pBBR1 replicon which allows for

replication in a number of different bacterial hosts [223]. It also includes a gentamicin

resistance cassette for selection of transformants, and a mobilization region for

conjugative transfer of plasmids into those species that are not amenable to

transformation through mechanical or chemical means.

pBBR-Dest42 was created through modification of the pBBR1MCS5 backbone to

include a Gateway cloning cassette so that pBBR-Dest42 can serve as a destination

vector for the Gateway cloning system (Invitrogen, Carlsbad, CA). The Gateway system

of recombination is based upon highly specific and efficient site-specific recombination

that integrates the bacteriophage lambda sequence into the genome of E. coli [225].

Vectors within the Gateway system contain bacteriophage lambda recombination sites

that have been modified to improve efficiency and specificity and to facilitate directional

cloning. Recombinase enzyme mixes are used for each step in the creation of both entry

and destination clones in order to take advantage of the recombination specificity of each

step. In the gateway system of cloning, an entry clone is created in two steps. First the

gene of interest is amplified through the polymerase chain reaction (PCR) using primers

210

that contain modified lambda phage attB attachment site sequences. This PCR product is

then cloned into lambda phage attP recombination sites of the entry vector through a BP

clonase reaction, creating an attL recombination site. The entry vector contains a

kanamycin resistance cassette for selection purposes. The gene of interest in the entry

vector is flanked by lambda phage attL attachment sites, and can be subcloned into a

destination vector containing attR recombination sites through an LR recombinase

reaction. The attR recombination sites flank a cloning cassette in the destination vector

that contains a chloramphenicol resistance gene and a ccdB lethality gene from the F

plasmid [225]. The ccdB gene product interferes with DNA gyrase, and thus destination

plasmids must be maintained in a strain of bacteria such as E. coli species DB3.1

(Invitrogen, Carlsbad CA) with a mutated DNA gyrase that is impervious to the effects of

CcdB [225]. Once an LR recombination reaction is done, this cassette is replaced with

the gene of interest, and the destination clone is then used for expression of fusion

protein. Inclusion of this gateway cloning cassette in the parental pBBR-Dest42 vectors

provides the advantages of a quick and easy ligation-independent cloning method [214,

234-238] that facilitates a high-throughput approach to both surveying protein-protein

interactions on a genome-wide scale and to rapidly expressing single fusion proteins of

interest.

The pBBR-Dest42 parent vector contains a C-terminal tag consisting of 6xHIS in

tandem with a V5 epitope tag and has been used to express tagged “bait” proteins in soil

bacterium R. palustris on a genome-wide scale [81]. Over 800 tagged “bait” proteins

have been expressed and purified in a multi-step purification strategy compatible with the

HIS-V5 tandem tag. The eluates were analyzed via mass spectrometry to identify bait

211

proteins and their associated interacting partners. While the 6xHIS tag is a commonly

used tag with a simple IMAC (Immobilized Metal Affinity Column) purification strategy,

elution of bound bait protein to Nickel-NTA via the 6xHIS and to the anti-V5 resin via

the V5 epitope tag requires harsh elution conditions, denaturing the bait protein and

possibly disrupting interactions of partner proteins or losing the interacting proteins

altogether.

Design considerations for inclusion of new tags in pBBR-Dest42-based vectors

It is sometimes desirable to purify proteins under non-denaturing native biological

conditions. Non-denaturing conditions used during purification can improve yields and

facilitate maintenance of non-covalent interactions between protein binding partners

[239, 240]. Further, the ability to purify a protein and associated interacting partners can

be dependent upon the tag or the combination of tags used for the purification, and upon

the position of the tag within the fusion protein [234]. Overlap of protein interacting

partners identified with different purification strategies can be as low as 4-7% [73].

Therefore, having a set of vectors containing three different C-terminal tandem affinity

tags with the backbone based on the pBBR-Dest42 parent vector can be useful in

isolating different complexes that may not be amenable to purification using a HIS-V5

tag, and can provide complementary strategies to increase the numbers of proteins

amenable to purification on a genome-wide scale. To address these issues, the C-

terminal HIS-V5 tag in the pBBR-Dest42 parent vector was replaced with either a

modified Tandem Affinity Purification (TAP –2xCalmodulin Binding Protein(CBP) -

Tobacco Etch Virus (TEV) protease site –2xProtein A) tag [80], a modified Sequential

Peptide Affinity (SPA- 2xCBP - TEV protease site - 3xFLAG epitope) tag [75] or a

212

StrepII-TEV protease site-6xHIS (STH) tag [233], as shown in Figure 5.3. Use of only

C-terminal fusion tags alleviated the need for including ribosomal binding sites that

would be required for an N-terminal tag (and thus would likely further limit the

application of the vectors) so for this initial work only C-terminal tags were included.

Use of tag systems which combine two different affinity tags, such as the TAP

tag developed by Rigaut et al. [80] or the SPA tag developed by Zeghouf et al [75] can

increase the applicability and flexibility of a purification scheme, resulting in a higher

number of proteins expressed due to increased solubility or increased yield resulting from

the combination of tags [241]. The TAP tag is a large tag of ~27kDa which has been

used in successful purifications of bait proteins and elucidation of protein interaction

networks in both yeast [76-79] and E. coli [73, 74], and for this reason was chosen for

placement in the pBBR-Dest42 backbone vectors. Larger tags have been shown to

increase the solubility and stability of proteins when expressed as fusion proteins, and

often can improve the ability to purify these proteins [239]. The SPA tag, which at ~9

kDa is smaller than the TAP tag, has been used for affinity purification of large numbers

of protein complexes in E. coli, with results comparable to those obtained using bait

proteins tagged with a TAP tag [74, 75, 83]. Since not all proteins will express with large

tags, the SPA tag was also chosen for inclusion in the pBBR-Dest42-based vectors. The

STH tag was chosen for inclusion due to its small size, inexpensive purification reagents,

and for its demonstrated success in isolating human protein complexes [233]. Since a

smaller tag has less likelihood of interference with protein folding, or with association of

interacting partners, the SPA and STH tags may allow purification of bait proteins and

associated binding partners that will not express with the larger TAP tag.

213

As described earlier, TAP tag sequences were obtained from Rigaut et al [80],

SPA tag sequences from Zeghouf et al [75] and STH tag sequences from Rich Gianonne

(personal communication). STH tag sequence was used as designed. Modifications to

the original TAP and SPA sequences included an additional consensus sequence for

Tobbaco Etch Virus (TEV) protease between the two affinity epitopes, and inclusion of a

tetracysteine (4C) tag with the sequence CCPGCC upstream of the tag sequence.

Protease cleavage is a commonly used strategy in purification schemes [75, 80, 231, 238,

241], allowing for removal of part or all of the tag, which is a huge advantage for

crystallization of proteins to determine structure [237] or for purifying proteins for

medical uses [242]. In a tandem affinity purification scheme, removal by proteolytic

cleavage of an outside tag that has very tight binding to affinity resins, such as Protein A

to IgG resins, allows for native elution of the protein in all steps [73-75, 80, 243]. As

explained in Chapter 1, the outer tag, tightly bound to the affinity resin, is used for initial

capture of the “bait” protein. This tag is then cleaved off by the protease, leaving “bait”

protein with only an inner tag floating free in solution. A second affinity capture is

performed using an inner tag with lower affinity binding that can then be eluted from the

affinity resin using gentler elution conditions. TEV protease, originally isolated from the

RNA genome of Tobacco Etch Virus (TEV), exhibits a high degree of cleavage

specificity [231, 244]. Recent improvements in the ability to purify it from a bacterial

host have made it an attractive choice for use in tandem affinity tags [217]. An

additional TEV cleavage site was added to the tag constructs inserted in the pBBR-

Dest42 backbone in order to improve likelihood and efficiency of cleavage [84, 233].

Improved efficiency results in 95% complete cleavage of inner tag from the protein

214

within 1 hour at room temperature (Rich Giannone, personal communication).

Tetracysteine (4C) tags were also included in new tag constructs in the innermost

position closest to the gene of interest. Tetracysteine (4C) tags have been used to

visualize protein localization in mammalian cells and to detect presence of expressed

protein within cell lysates [84]. Protein presence within cell lysates is visualized through

interaction of the 4C motif with TC-FlAsH (Fluorescein Arsenical Hairpin) reagent

(Molecular Probes, a division of Invitrogen, Carlsbad, CA). Reaction of the tag with the

FlAsH reagent creates a hairpin structure with the tag binding to arsenic residues in the

reagent. The dye then fluoresces with an emission wavelength of 550nm, allowing easy

detection in a fluorescent imager or transilluminator equipped with a 500-600nm filter, or

alternatively using an ethidium bromide channel. The dye is sensitive enough to detect

nanogram quantities of protein. Presence of the 4C motif in the tags allows detection of

expressed protein in cell lysates in the time it takes to run an SDS-PAGE gel, thus

eliminating the need for more time-consuming western blot procedures to determine

protein expression.

When isolating proteins and their interacting partners via an affinity purification

scheme with subsequent identification of those proteins via mass spectrometry (AP-MS),

the presence of non-specific interacting proteins becomes an issue [9]. Proteins such as

ribosomal proteins and elongation factors are simply “sticky,” adhering to resins and

attached protein complexes no matter what purification scheme is used. Tandem affinity

purification schemes serve to decrease the numbers of non-specifically interacting

proteins [80]. A number of methods have been developed to increase the probability that

detected proteins are true interacting partners [9, 12] but a survey of such methods is

215

beyond the scope of this work. Instead, the focus here is on expression of proteins.

Proteins expressed from a plasmid can vary greatly in expression level, depending

on a number of factors such as promoters used for protein expression and copy number of

the plasmid. Earlier studies in R. palustris have shown expression levels of RNA

polymerase subunits expressed from the lac promoter on a plasmid to vary between 0.7

and 1.7 times that of wild-type expression [11]. Over-expression of a protein can lead to

artifactual interactions that are not biologically relevant. Further, expression of tagged

versions of protein off of a plasmid increases the pool of the expressed protein in the cell,

with both a wild-type and a tagged version of the same protein being present in the cell.

This increases the competition for binding partners, and may lead to an inability to detect

interacting partners or an increase in the numbers of non-specific binding to the tagged

protein.

For this reason, a set of vectors based on the pJQ200KS plasmid was created that

would allow endogenous expression of tagged versions of proteins. pJQ200KS is a small

plasmid derived from the P15A plasmid first isolated from E. coli [226]. It contains a

gentamicin resistance cassette for selection of positive transformants, a mobilization

region for conjugative transfer, and can be maintained in E. coli. However, the P15A

origin of replication (ORI) ensures nonreplication in non-enterobacteria, and will

integrate into the genome of non-enterobacteria through homologous recombination [226,

227]. Gene replacement through use of pJQ200KS was first tested in Rhizobium

leguminosarum species and applied to many bacterial species since then, so success in the

use of this plasmid for gene integration in diverse bacteria has already been established

[226]. Usefulness of this vector in protein-protein interaction studies in non-

216

enterobacteria could include expression of tagged versions of proteins that are lethal,

either disrupting cellular function or killing the cell if over-expressed, or expressing

versions of proteins that will not express off of an exogenous plasmid, or expression of

low expression level proteins such as DNA polymerase subunits which are difficult to

purify with associated interacting partners due to competition with endogenously

expressed subunits for binding partners. Since this vector integrates into the genome of

non-enterobacteria, it is possible that integration will disrupt the expression of an

essential protein and thus be lethal to the cell. An additional disadvantage to the use of

an integrating vector containing C-terminal tags include limiting the genes expressed to

those that are either orphans within the genome or at the end of an operon structure.

Since no promoter sequence is included in the vector, C-terminally tagged proteins will

express from a native promoter in the genome sequence, but expression of those genes

downstream of the tagged protein may be disrupted. Endogenous expression under native

promoter is less likely to yield false positives because it limits somewhat the recombinant

protein expression levels and thus possible artifacts associated with over-expression.

Integration of tagged protein in genome disrupts native copy, and thus expression of a

tagged version under a native promoter could allow exploration of protein complexes that

will not purify with their interacting partners due to competition with non-tagged protein

subunits.

Construction of vectors with new tag sequences

Initial strategies for construction of new vectors included amplification of TAP2,

SPA2 and STH tag sequences designed and constructed earlier for inclusion in

mammalian expression vectors [233]. PCR product obtained from these vectors was

217

restriction digested and attempts were made to ligate the digested PCR product into

linearized pBBR-Dest42 as described above. After many attempts, this approach proved

unsuccessful. New tag sequence was designed from back translations of tag amino acid

sequences, with codons being optimized for expression in R. palustris as described

above. New tag sequences were constructed by GenScript (Piscatawny, NJ), and

constructs were amplified through PCR. A PCR amplification strategy was found to be

necessary due to the methylation sensitivity of the Cla1 restriction enzyme. Vectors

propagated through growth in E. coli strains contain methylated DNA which could not be

restriction digested by Cla1 endonuclease. A PCR strategy amplified the tag sequence

without methylation at the restriction enzyme consensus sequence. The pBBR-Dest42

parent vector was designed with restriction digest sites in two places: 1. surrounding the

original HIS-V5 tag for easy tag removal and replacement and 2. placed at the 5’ end of

the Gateway cloning cassette for inclusion of cassettes with N-terminal tags and

ribosomal binding sites. Parent vector was linearized and HIS-V5 tags removed through

digestion with BstB1 restriction endonuclease, which leaves a complementary

overhanging sequence to Cla1 restriction digest. PCR product for each individual tag

(STH, SPA2 or TAP2) was restriction digested with Cla1 restriction endonuclease and

ligated into BstB1 restriction sites of the linearized parent vector. Ligation of Cla1

digested tags into the parent vector BstB1 restriction sites destroyed the BstB1 restriction

site. Therefore, restriction digest of the ligated vectors with BstB1 restriction

endonuclease provided a convenient way to eliminate vectors that had re-ligated without

a tag insert or had re-ligated with the original HIS-V5 vector. All constructs were

confirmed both with diagnostic restriction digest and with DNA sequencing. This same

218

strategy can easily be used to create vectors with additional C-terminal or with N-

terminal tags with the pBBR-Dest42 backbone .

The above strategy was also used to insert C-terminal tags into pJQ200KS

parent vector [226]. The gateway cassettes including either TAP2, SPA2 or STH C-

terminal tags were PCR amplified using universal primers. PCR product was then ligated

into a linearized pJQ200KS parent vector, as described above.

Protein expression

Rhodopseudomonas palustris was used as a bacterial host to show protein

expression and to illustrate the usefulness of both integrating and exogenous vectors in

investigation of protein complexes in a native host. These bacteria were chosen for the

proof-of concept set of experiments because of the availability at ORNL of a large

amount of data compiled on both a proteome expression level and a functional

proteomics level in these bacteria, thereby providing a suite of reference data [30, 81].

The pBBR-Dest42 parent vector from which the new set of vectors was created has been

tested extensively in this species, thus facilitating comparison of expression of proteins

with new tags to those already characterized. Additionally, the availability of a wide

variety of entry clones for native bacterial proteins in this species made it an attractive

host for the protein-protein interaction studies in this chapter. R. palustris is also an

attractive host for testing protein expression from the integrating vector set based on the

pJQ200KS parent vector due to previous use of this parent vector to create gene knock-

outs in this species [245, 246]. Perhaps the largest disadvantage to using R. palustris as a

bacterial host for protein expression lay in the amount of time required for growth, with

approximately 6 weeks being required from transformation to initial expression tests.

219

The original goals of this project included not only creation of vectors with

different C-terminal tags such that coverage of protein-protein interactions could be

increased, but also comparative tests of protein expression across tags. The questions

that we hoped to address initially included examining the effects of endogenous

expression versus exogenous expression on detection of proteins and their interacting

partners. Further, we hoped to characterize the effects of different tags on expression and

immunoprecipitation of proteins of low versus high expression level, and the effect of

different tag sizes on expression and detection of proteins of high versus low molecular

weight. We also hoped to examine the usefulness of any of the new C-terminal tags in

the detection of binding partners of transiently interacting proteins. We further hoped to

confirm the use of the 4C tag for detection of expression within cell lysates.

Original candidate test gene choices for pBBR-Dest42-based vectors included 2

subunits of DNA polymerase, rpa_0301 DNA polymerase epsilon chain, dnaQ, and

rpa_0615 DNA polymerase tau subunit, dnaX, (both with low expression levels),

rpa_1175 CheY-like protein (transient interactor) and 2 subunits of RNA polymerase,

rpa_3226 RNA polymerase alpha subunit, rpoA (high expression level, low molecular

weight) and rpa_3267 RNA polymerase beta prime subunit, rpoC (high expression level,

high molecular weight). Unfortunately, due to the time required to create the new vectors

combined with the long growth time for R. palustris, RNA polymerase subunits rpa_3226

(RpoA) and rpa_3267 (RpoC) were the only proteins successfully expressed, affinity

purified and detected via mass spectrometry. Although all other proteins showed some

level of expression in cell lysates analyzed by western blot, purification and subsequent

detection of the tagged bait proteins and interacting partners in mass spectrometry

220

experiments proved unsuccessful. Original gene choices for expression in pJQ200-KS-

SPA2 vector included rpa_3226 rpoA (high expression level), rpa_3059 HolC subunit of

DNA polymerase (low expression level), rpa_3522 FtsZ cell division protein, rpa_1631

CheW scaffold protein (no expression off of an exogenous plasmid), and rpa_1175

CheY-like signal protein. Each of these genes was found either at the end of an operon or

as an orphan gene not found in an operon structure. Again, although expression in cell

lysates was confirmed via western blot, only the RpoA subunit of RNA polymerase was

isolated and purified, with bait protein and binding partners detected through mass

spectrometry. Perhaps with more time available, immunoprecipitation parameters, such

as salt and detergent content of buffers and amount of resins used, and incubation times

could have been optimized such that low expression level proteins could have been

detected.

RNA polymerase is a multi-subunit enzyme that performs the transcription of

DNA template into an RNA polymer. It is well-characterized and abundant in the cell ,

and has been extensively studied in E. coli and R. palustris [11, 74, 81]. The core of

bacterial RNA polymerase is composed of 5 subunits, consisting of 2 alpha subunits

(RpoA, 36.5 kDa), 1 beta subunit (RpoB, 150.6 kDa), 1 beta prime subunit (RpoC, 155.2

kDa), and 1 omega subunit (RpoZ, 10.2 kDa) [11, 247]. Specificity of RNA polymerase

is conferred through association of the RNA polymerase core with various sigma factors,

which recognize and bind tightly to gene or operon promoter sequences. Different sigma

factors are employed at different times, especially during stress conditions, thus allowing

for adaptability in response to environmental stress [141]. In earlier genome-wide

protein interaction studies conducted in R palustris, RPA1288 RpoD sigma factor was

221

found to interact with both RpoA and RpoC. Additional binding partners for RNA

polymerase alpha subunit (RpoA) and for beta prime subunit (RpoC) are cataloged in the

STRING database (http://string.embl.de/ newstring_cgi/), and representative figures of

interacting partners are shown in Figure 5.4. Because earlier experiments resulted in a

thorough catalog of interacting partners for this model system, RNA polymerase alpha

(RPA3226, RpoA) and beta prime (RPA3267, RpoC) subunits served as useful bait

proteins for verifying expression of low and high molecular weight fusion protein,

respectively, from newly created vectors. Fusion proteins of RpoA with either STH tag

or SPA2 tag or TAP2 tag were expressed, immunoprecipitation experiments performed

and results tabulated in Table 5.4. RpoC was expressed as a fusion protein with either

STH tag or with SPA2 tag, but attempts to express a RpoC-TAP2 fusion protein were

unsuccessful. Results of immunoprecipitation experiments with RpoC are tabulated in

Table 5.5. Thus, the efficacy of this vector set has been shown in initial tests.

Conclusions

The creation of a set of Gateway®

compatible destination vectors based on the

parent vector pBBR-Dest42, gives an advantage of protein expression in a broad range of

bacterial hosts. However, as demonstrated above in a set of preliminary experiments,

while the constructs are validated, their application should be tested further, perhaps by

using other host strains (such as Azospirillum) and/or other bait proteins. These vectors

can potentially be used for exploration of protein complexes expressed in native hosts,

222

Figure 5. 4 Schematic depiction of RNA polymerase subunit interacting partners

(http://string.embl.de)

Figures were obtained from the STRING interaction database (http://string.embl.de/ newstring_cgi/)A)

Interacting partners for RpoA (RPA3226), B) Interacting partners for RpoC (RPA3267). Each “ball”

represents a different protein, while each line between individual proteins represents a different form of

evidence of interaction, either experimental or in silico. Stronger evidence of interactions between two

proteins is represented in this view by more lines between the two proteins.

B) RpoC (RPA3267) interacting partners

A) RpoA (RPA3226) interacting partners

223

and are amenable to high-throughput but their application needs to be confirmed. Further

characterization of these new vectors could include comparison tests of high versus low

expression level proteins with each tag after optimization of affinity purification

protocols. Additional tests could include a test of detection of protein presence through

using the 4C epitope, which could prove much easier in a bacterial species that does not

require chemical lysis procedures.

224

Chapter 6. Conclusion

Mass spectrometry based proteomics is a powerful tool to use in the

characterization of protein expression of newly sequenced microbes. Although genome

sequence is an important indication of the physiological capability of a microbe, the

expressed proteome gives a more complete picture of the actual physiology of the cells

under given conditions. Many microbial genomes contain multiple copies of a number of

genes, and genome sequence alone can not tell you which one is expressed in a given

situation. Proteomics of bacterial cells grown under different conditions can provide a

picture, a snapshot in time, of those proteins expressed and those that are important to

existence within a given environmental condition or a given growth state. Thus proteomic

analysis can at the very least complement an informed genome sequence annotation.

In Chapter 3 of this work, the tool of mass spectrometry-based proteomics was

applied to investigate the expression proteome of single isolate cultures of newly

sequenced soil diazotrophic bacteria, Azospirillum brasilense. The proteomes of two

different strains of A. brasilense were examined after growth under optimal conditions in

minimalmedia either containing nitrogen in the form of ammonium sulfate, or not

containing a nitrogen source and grown under limited oxygen conditions to promote

nitrogen fixation (nitrogen fixing conditions). These two cultures were grown under two

well-characterized optimum growth conditions, corresponding to distinct physiological

abilities for this species. Since no genome sequence is yet available for strain Sp7, the

genome of closely related Sp245 strain was used in this work. BlastX comparison of the

available Sp7 translated coding sequences with those of the Sp245 strain gene coding

sequences indiciated a high degree of similarity (90-100%) between individual coding

225

sequences of each strain, although one gene (Sp7 nirS) did not have a homolog in the

Sp245 genome, and two other coding sequences (nitrogenase alpha and beta chains)

exhibited only ~30% similarity to those in the Sp245 genome. Therefore, inferences

regarding protein identification in strain Sp7 using the annotated genome sequence of

strain Sp245, while informative, is likely to have some flaws. A recent study indicated

that the wzm gene found on the pRHICO plasmid of Sp7 strain has 98% identity to the

same gene found within the Sp245 genome [172]. Although some genes may have DNA

sequence differences, BLASTx comparison of available Sp7 genes suggests that the

differences reflected at a protein level seem to be minimal. Nevertheless, some

artifactual identifications could be made, or some proteins not detected due to sequence

differences. Therefore, the genome sequence for the Sp245 strain is likely the best

representation of genome sequence (and therefore translated protein sequence) available

with the majority of proteins detected being accurate. However, due to the possible

sequence differences between strains, direct comparisons between expression proteomes

of each strain could not be done with sufficient confidence to be insightful and accurate.

The high throughput methodology applied to Azospirillum brasilense has yielded

the first proteome results of this metabolically diverse bacterial species, providing an

initial foundation for further in-depth analyses. Combination of proteomic data with

microarray studies currently being conducted will further elucidate the overall changes in

Azospirillum cells due to nitrogen fixation, giving a more complete picture of overall

physiological changes. Combination of these two “omics” technologies could further

provide a set of useful markers for physiological changes occurring as the cells fix

nitrogen or as they prepare to fix nitrogen. This could in turn lead to development of

226

reporter genes useful for determining physiological changes related to nitrogen fixation,

in addition to providing insight into the physiology of cells grown under these conditions.

Additionally, the proteome data can be useful in developing a more complete annotation

of genome sequences. Searching proteome data using a a six frame translation of the

genome sequence using a methodology such as that used on data from the plant

pathogens Phytopthera sojae and Phytopthera ramorum can yield more accurate start and

stop sites for some genes and can lead to a more accurate annotation of putative genes.

This method was used on, and successfully used to provide more complete annotation to

the genome sequence of these plant pathogens [248]. Future directions could include a

six-frame translation of the Sp245 proteome data such that genome annotation could be

more complete.

An additional expression proteome study was done to investigate the effect of a

chemotaxis-like signaling pathway on the overall physiology of Azospirillum brasilense

species Sp7, with results presented in Chapter 4. Expression proteomes of cultures of

mutant cells were done in order to elucidate the overall proteome changes when forward

signaling output was eliminated versus when only the adaptation pathway was

eliminated. The broader view of the physiology of the mutant cells provided by the

proteomic study told a story of a dramatic difference in physiology between the two

isogenic mutants. There was not a substantial difference between the wild-type cells and

the cells which lacked an adaptation pathway, but the differences noted between wild-

type and those cells lacking a forward signaling pathway was much different than was

previously known. Although a common set of proteins was expressed, representing a

number of proteins needed for basic cellular function, the patterns of expression levels of

227

these proteins was very different in the two mutants. Further analysis of proteomic data

could provide a discovery of useful markers for physiological studies, perhaps providing

a set of useful reporter genes that could indicate certain physiological features such as

those occurring in preparation for nitrogen fixation. In addition, given that the molecular

mechanisms underlying the different behaviors of the two mutant strains compared is yet

to be determined, this comparison lays the foundation for identifying candidate functions

that are directly or indirectly modulated by the chemotaxis-like pathway.

Perhaps the most interesting finding was the expression of type 6 secretion system

(T6SS) structural components, which mediates interaction with a eukaryotic host, found

to be important in nodulation in Rhizobium, and in long-term survival in chronic

infections of pathogenic bacteria. Found to be expressed and upregulated in forward

signaling mutants, many components were not detected in adaptation mutants, suggesting

some level of importance in the response to the Che1 signaling pathway. Since little is

known about T6SS in non-pathogenic bacteria, these results are intriguing, and could lead

to a number of further investigations. For instance, whether expression of T6SS

components is regulated directly or indirectly by the Che1 operon or whether it results

from the difference in perception of the environment (as evidenced by the distinct

physiologies of the cells brought to light by the proteomic analysis) is yet to be

determined. Further, the signal input to which this signaling system is responding is yet

to be elucidated. All data was searched using a +14 Da modification on glutamate

residues, indicative of addition of a single methyl group, although this analysis was not

included in Chapter 4. Sensory receptors involved in signaling environmental conditions

through chemotaxis pathways are often constitutively modified by methylation, with

228

adaptive response including removal of methylation. Further investigation of methylation

patterns within this data set could provide a clue as to what receptors are methylated in

response to the Che1 chemotaxis-like pathway. Little is known about the structure of the

T6SS in non-pathogenic bacteria, but structural components in pathogenic bacteria are

assumed to consist of both an Hcp protein and a Vgr-family protein that combine to form

the secretory apparatus. Since there is no Hcp protein within the genome of Sp245

species, an investigation of protein binding partners for the T6SS components detected in

the proteome study could provide clues as to the construction of the secretory apparatus

for these cells. Further, the function of the T6SS in mediating cellular changes related to

aggregation, and possible secreted effectors that may mediate interaction of the

Azospirillum cells with plant roots could be investigated.

Finally, in Chapter 5, a set of broad host range expression vectors, and an

additional set of integrating vectors with a cassette conferring Gateway cloning

compatibility was constructed. These vectors contain DNA sequences for different C-

terminal tandem affinity tags. Since proteins are not homogeneous molecules, affinity

tags lend the capability to do generic purifications, such that proteins can be purified in a

high-throughput manner. A set of vectors allows experimentation such that a wider

variety of proteins can possibly be expressed. Gateway compatibility further simplifies

cloning such that a large percentage of the genome could be expressed and purified and

interacting partners determined, thus entering the realm of functional proteomics.

Expression of tagged RNA polymerase subunits from these vectors was tested in R.

palustris, a soil bacterium in which the base parent vector had been extensively tested

and for which a thorough set of entry clones were already constructed. Future directions

229

could include a more thorough characterization of these vectors as outlined in Chapter 5.

Further, expression of proteins for these vectors could be tested in Azospirillum species

with a model complex such as RNA polymerase. After successful tests in Azospirillum,

these vectors could be used to express and purify the components of the T6SS and

interacting partners identified by mass spectrometry, perhaps facilitating a model of

secretory apparatus.

In summary, the expression proteomics presented in this work yields the first set

of proteomics data for Azospirillum brasilense. This data set expands the base of

knowledge available on this versatile bacterium, and provides a set of proteins for further

exploration. Further, the set of broad host range vectors developed provide a set of tools

to use for exploration of protein interactions on a genome-wide scale that could lead to a

functional understanding of individual proteins.

230

References

231

1. Kocher, T. and G. Superti-Furga, Mass spectrometry-based functional

proteomics: from molecular machines to protein networks. Nature Methods, 2007.

4(10): p. 807-815.

2. McDonald, W.H. and J.R. Yates III, Shotgun proteomics and biomarker

discovery. Disease Markers, 2002. 18(2): p. 99.

3. Eng, J.K., A.L. McCormack, and I. John R. Yates, An Approach to Correlate

Tandem Mass Spectral Data of Peptides with Amino Acid Sequences in a Protein

Database. Journal of the American Society of Mass Spectrometry, 1994. 5: p.

976-989.

4. Han, X., A. Aslanian, and J.R. Yates III, Mass spectrometry for proteomics.

Current Opinion in Chemical Biology, 2008. 12(5): p. 483.

5. MacCoss, M.J. and John R. Yates III, Proteomics: analytical tools and

techniques. Current Opinion in Clinical Nutrition and Metabolic Care, 2001. 4: p.

369-375.

6. Tabb, D.L., C.G. Fernando, and M.C. Chambers, MyriMatch: Highly Accurate

Tandem Mass Spectral Peptide Identification by Multivariate Hypergeometric

Analysis. Journal of Proteome Research, 2007. 6(2): p. 654.

7. Tabb, D.L., et al., DBDigger: Reorganized Proteomic Database Identification

That Improves Flexibility and Speed. Analytical Chemistry, 2005. 77(8): p. 2464-

2474.

8. Tabb, D.L., A. Saraf, and J.R. Yates, GutenTag: High-Throughput Sequence

Tagging via an Empirically Derived Fragmentation Model. Analytical Chemistry,

2003. 75(23): p. 6415.

9. Tackett, A.J., et al., I-DIRT, A General Method for Distinguishing between

Specific and Nonspecific Protein Interactions. Journal of Proteome Research,

2005. 4: p. 1752-1756.

10. Ong, S.-E., et al., Stable Isotope Labeling by Amino Acids in Cell Culture, SILAC,

as a Simple and Accurate Approach to Expression Proteomics. Mol Cell

Proteomics, 2002. 1(5): p. 376-386.

11. Hervey, W.J., et al., Evaluation of Affinity-Tagged Protein Expression Strategies

Using Local and Global Isotope Ratio Measurements. Journal of Proteome

Research, 2009. 8(7): p. 3675.

12. Lee, D.J., et al., Affinity Isolation and I-DIRT Mass Spectrometric Analysis of the

Escherichia coli O157:H7 Sakai RNA Polymerase Complex. Journal of

Bacteriology, 2008. 190(4): p. 1284-1289.

13. Thompson, M.R., et al., Experimental Approach for Deep Proteome

Measurements from Small-Scale Microbial Biomass Samples. Analytical

Chemistry, 2008. 80(24): p. 9517.

14. McCormack, A.L., et al., Direct Analysis and Identification of Proteins in

Mixtures by LC/MS/MS and Database Searching at the Low-Femtomole Level.

Analytical Chemistry, 1997. 69(4): p. 767.

232

15. Liu, K., et al., Relationship between Sample Loading Amount and Peptide

Identification and Its Effects on Quantitative Proteomics. Analytical Chemistry,

2009. 81(4): p. 1307.

16. Link, A.J., et al., Direct analysis of protein complexes using mass spectrometry.

Nature Biotechnology, 1999. 17.

17. McDonald, W.H., et al., Comparison of three directly coupled HPLC MS/MS

strategies for identification of proteins from complex mixtures: single-dimension

LC-MS/MS, 2-phase MudPIT, and 3-phase MudPIT. International Journal of Mass

Spectrometry, 2002. 219: p. 245-251.

18. Peng, J., et al., Evaluation of Multidimensional Chromatography Coupled with

Tandem Mass Spectrometry (LC/LC-MS/MS) for Large-Scale Protein Analysis:

The Yeast Proteome. Journal of Proteome Research, 2003. 2(1): p. 43-50.

19. Washburn, M.P., D. Wolters, and J.R.Yates III, Large-scale analysis of the yeast

proteome by multidimensional protein identification technology. Nature

Biotechnology, 2001. 19(March 2001): p. 242-247.

20. Wolters, D.A., M.P. Washburn, and John R Yates III, An Automated

Multidimensional Protein Identification Technology for Shotgun Proteomics.

Analytical Chemistry, 2001. 73: p. 5683-5690.

21. Bell, A.W., et al., A HUPO test sample study reveals common problems in mass

spectrometry-based proteomics. Nat Meth, 2009. 6(6): p. 423.

22. Prenni, J.E., A.C. Avery, and C.S. Olver, Proteomics: a review and an example

using the reticulocyte membrane proteome. Veterinary Clinical Pathology, 2007.

36(1): p. 13-24.

23. Wu, C. and I. John R. Yates, The Application of Mass Spectrometry to Membrane

Proteins. Nature Biotechnology, 2003. 21: p. 262-267.

24. Zybailov, B., et al., Statistical Analysis of Membrane Proteome Expression

Changes in Saccharomyces cerevisiae. Journal of Proteome Research, 2006. 5(9):

p. 2339-2347.

25. Tebbe, A., et al., Analysis of the cytosolic proteome of Halobacterium salinarum

and its implication for genome annotation. Proteomics, 2005. 5: p. 168-179.

26. Djordjevic, M.A., Sinorhizobium meliloti metabolism in the root nodule: A

proteomic perspective. Proteomics, 2004. 4(7): p. 1859-1872.

27. Koksharova, O.A., J. Klint, and U. Rasmussen, Comparative proteomics of cell

division mutants and wild-type of Synechococcus sp. strain PCC 7942.

Microbiology, 2007. 153(8): p. 2505-2517.

28. Malmstrom, J., et al., Proteome-wide cellular protein concentrations of the

human pathogen Leptospira interrogans. Nature, 2009. 460(7256): p. 762.

29. Ow, S.Y., et al., Quantitative Overview of N2 Fixation in Nostoc punctiforme

ATCC 29133 through Cellular Enrichments and iTRAQ Shotgun Proteomics.

Journal of Proteome Research, 2009. 8(1): p. 187-198.

30. VerBerkmoes, N.C., et al., Determination and Comparison of the Baseline

Proteomes of the Versatile Microbe Rhodopseudomonas palustris under Its Major

Metabolic States. J. Proteome Res., 2006. 5(2): p. 287-298.

31. Victor, J.N. and J.F.S. Marc, Analysis of environmental stress response on the

proteome level. Mass Spectrometry Reviews, 2008. 27(6): p. 556-574.

233

32. Wagner, M.A., et al., Global analysis of the Brucella melitensis proteome:

Identification of proteins expressed in laboratory-grown culture. Proteomics,

2002. 2(8): p. 1047-1060.

33. Yang-Hoon, K., et al., Proteome response of Escherichia coli fed-batch culture to

temperature downshift. Applied Microbiology & Biotechnology, 2005. 68(6): p.

786.

34. Moberg, M., J. Bergquist, and D. Bylund, A generic stepwise optimization

strategy for liquid chromatography electrospray ionization tandem mass

spectrometry methods. Journal of Mass Spectrometry, 2006. 41(10): p. 1334-

1345.

35. Zybailov, B., et al., Correlation of Relative Abundance Ratios Derived from

Peptide Ion Chromatograms and Spectrum Counting for Quantitative Proteomic

Analysis Using Stable Isotope Labeling. Analytical Chemistry, 2005. 77(19): p.

6218-6224.

36. Timms, J.F. and R. Cramer, Difference gel electrophoresis. Proteomics, 2008.

8(23-24): p. 4886-4897.

37. Mottaz-Brewer, H.M., et al., Optimization of Proteomic Sample Preparation

Procedures for Comprehansive Protein Characterization of Pathogenic Systems.

Journal of Biomolecular Techniques, 2008. 19: p. 285-295.

38. Wang, H. and S. Hanash, Intact-protein based sample preparation strategies for

proteome analysis in combination with mass spectrometry. Mass Spectrometry

Reviews, 2005. 24(3): p. 413-426.

39. Tang, J., et al., Recent development of multi-dimensional chromatography

strategies in proteome research. Journal of Chromatography B, 2008. 866(1-2): p.

123.

40. Domon, B. and R. Aebersold, Mass Spectrometry and Protein Analysis. Science,

2006. 312(5771): p. 212-217.

41. John R Yates III, Mass spectrometry and the age of the proteome. Journal of Mass

Spectrometry, 1998. 33(1): p. 1-19.

42. Khalsa-Moyers, G. and W. Hayes McDonald, Developments in mass spectrometry

for the analysis of complex protein mixtures. Briefings in Functional Genomics &

Proteomics, 2006. 5(2): p. 98.

43. McDonald, W. and J.R.Y. III, Proteomic Tools for Cell Biology. Traffic, 2000. 1:

p. 747-754.

44. Karas, M. and F. Hillenkamp, Laser desorption ionization of proteins with

molecular masses exceeding 10,000 daltons. Analytical Chemistry, 1988. 60(20):

p. 2299.

45. Fenn, J.B., et al., Electrospray Ionization for Mass Spectrometry of Large

Biomolecules. Science, 1989. 246(4926): p. 64.

46. March, R.E., An Introduction to Quadrupole Ion Trap Mass Spectrometry.

Journal of Mass Spectrometry, 1997. 32: p. 351-369.

47. March, R.E., Quadrupole ion traps. Mass Spectrometry Reviews, 2009. March

2009.

234

48. Schwartz, J.C., M.W. Senko, and J.E.P. Syka, A two-dimensional quadrupole ion

trap mass spectrometer. Journal of the American Society for Mass Spectrometry,

2002. 13(6): p. 659.

49. Shukla, A.K. and J.H. Futrell, Tandem mass spectrometry: dissociation of ions by

collisional activation. Journal of Mass Spectrometry, 2000. 35: p. 1069-1090.

50. Tabb, D.L., et al., DirecTag: Accurate Sequence Tags from Peptide MS/MS

through Statistical Scoring. Journal of Proteome Research, 2008. 7(9): p. 3838.

51. Taylor, J.A. and R.S. Johnson, Implementation and Uses of Automated de Novo

Peptide Sequencing by Tandem Mass Spectrometry. Analytical Chemistry, 2001.

73(11): p. 2594.

52. Altschul, S.F., et al., Protein database searches using compositionally adjusted

substitution matrices. FEBS Journal, 2005. 272(20): p. 5101-5109.

53. Perkins, D.N., et al., Probability-based protein identification by searching

sequence databases using mass spectrometry data. Electrophoresis, 1999. 20(18):

p. 3551-3567.

54. Tabb, D.L., W.H. McDonald, and I. John R. Yates, DTASelect and Contrast:

Tools for Assembling and Comparing Protein Identifications from Shotgun

Proteomics. Journal of Proteome Research, 2002. 1: p. 21-26.

55. Albrethsen, J., Reproducibility in Protein Profiling by MALDI-TOF Mass

Spectrometry. Clin Chem, 2007. 53(5): p. 852-858.

56. Mayya, V., et al., Systematic Comparison of a Two-dimensional Ion Trap and a

Three-dimensional Ion Trap Mass Spectrometer in Proteomics. Molecular and

Cellular Proteomics, 2005. 4: p. 214-223.

57. Moberg, M., K.E. Markides, and D. Bylund, Multi-parameter investigation of

tandem mass spectrometry in a linear ion trap using response surface modelling.

Journal of Mass Spectrometry, 2005. 40: p. 317-324.

58. Riter, L.S., et al., Statistical design of experiments as a tool in mass spectrometry.

Journal of Mass Spectrometry, 2005. 40: p. 565-579.

59. Venable, J.D. and I. John R Yates, Impact of Ion Trap Tandem Mass Spectra

Variability on the Identification of Peptides. Analytical Chemistry, 2004. 76: p.

2928-2937.

60. Wenner, B.R. and B.C. Lynn, Factors that Affect Ion Trap Data-Dependent

MS/MS in Proteomics. Journal of the American Society of Mass Spectrometry,

2004. 15: p. 150-157.

61. Xie, H. and T.J. Griffin, Trade-Off between High Sensitivity and Increased

Potential for False Positive Peptide Sequence Matches Using a Two-Dimensional

Linear Ion Trap for Tandem Mass Spectrometry-Based Proteomics. Journal of

Proteome Research, 2006. 5: p. 1003-1009.

62. Jiang, X., et al., Optimization of filtering criterion for SEQUEST database

searching to improve proteome coverage in shotgun proteomics. BMC

Bioinformatics, 2007. 8: p. 323-334.

63. Gade, D., et al., Towards the proteome of the marine bacterium Rhodopirellula

baltica: Mapping the soluble proteins. Proteomics, 2005. 5(14): p. 3654-3671.

235

64. Maillet, I., et al., From the genome sequence to the proteome and back:

Evaluation of E. coli genome annotation with a 2-D gel-based proteomics

approach. Proteomics, 2007. 7(7): p. 1097-1106.

65. Zhang, M.-J., et al., Comparative proteomic analysis of passaged Helicobacter

pylori. Journal of Basic Microbiology, 2009. 9999(9999): p. NA.

66. Ha, D.-J., et al., Proteome analysis of Halobacterium salinarum and

characterization of proteins related to the degradation of isopropyl alcohol.

Biochimica et Biophysica Acta, 2007. 1774: p. 44-50.

67. Gevaert, K., et al., Stable isotopic labeling in proteomics. Proteomics, 2008. 8(23-

24): p. 4873-4885.

68. Wiese, S., et al., Protein labeling by iTRAQ: A new tool for quantitative mass

spectrometry in proteome research. PROTEOMICS, 2007. 7(3): p. 340-350.

69. Stevenson, S.E., et al., Validation of gel-free, label-free quantitative proteomics

approaches: Applications for seed allergen profiling. Journal of Proteomics,2009.

72(3): p 555-566.

70. Old, W.M., et al., Comparison of Label-free Methods for Quantifying Human

Proteins by Shotgun Proteomics. Mol Cell Proteomics, 2005. 4(10): p. 1487-

1502.

71. Girard, M., et al., Non-stoichiometric Relationship between Clathrin Heavy and

Light Chains Revealed by Quantitative Comparative Proteomics of Clathrin-

coated Vesicles from Brain and Liver. Mol Cell Proteomics, 2005. 4(8): p. 1145-

1154.

72. Florens, L., et al., Analyzing chromatin remodeling complexes using shotgun

proteomics and normalized spectral abundance factors. Methods, 2006. 40: p.

303-311.

73. Arifuzzaman, M., et al., Large-scale identification of protein-protein interaction

of Escherichia coli K-12. Genome Res., 2006. 16(5): p. 686-691.

74. Butland, G., et al., Interaction network containing conserved and essential protein

complexes in Escherichia coli. Nature, 2005. 433: p. 531-537.

75. Zeghouf, M., et al., Sequential Peptide Affinity (SPA) System for the Identification

of Mammalian and Bacterial Protein Complexes. Journal of Proteome Research,

2004. 3: p. 463-468.

76. Babu, M., et al., Systematic Characterization of the Protein Interaction Network

and Protein Complexes in Saccharomyces cerevisiae using Tandem Affinity

Purification and Mass Spectrometry, in Methods in Molecular Biology, I.

Staglijar, Editor. 2009, Humana Press. p. 187-207.

77. Gavin, A.-C., et al., Functional organization of the yeast proteome by systematic

analysis of protein complexes. Nature, 2002. 415(6868): p. 141.

78. Graumann, J., et al., Applicability of Tandem Affinity Purification MudPIT to

Pathway Proteomics in Yeast. Molecular and Cellular Proteomics, 2004. 3: p.

226-237.

79. Ho, Y., et al., Systematic identification of protein complexes in Saccharomyces

cerevisiae by mass spectrometry. Nature, 2002. 415(6868): p. 180.

236

80. Rigaut, G., et al., A generic protein purification method for protein complex

characterization and proteome exploration. Nature Biotechnology, 1999. 17: p.

1030-1032.

81. Pelletier, D.A., et al., A General System for Studying Protein-Protein Interactions

in Gram-Negative Bacteria. Journal of Proteome Research, 2008. 7(8): p. 3319-

3328.

82. Collins, M.O. and J.S. Choudhary, Mapping multiprotein complexes by affinity

purification and mass spectrometry. Current Opinion in Biotechnology, 2008.

19(4): p. 324.

83. Babu, M., et al., Sequential Peptide Affinity Purification System for the Systematic

Isolation and Identification of Protein Complexes from Escherichia Coli, in

Methods in Molecular Biology, J. Reinders and A. Sickmann, Editors. 2009,

Humana Press. p. 373-400.

84. Giannone, R.J., Y. Liu, and Y. Wang, The Monitoring and Affinity Purification of

Proteins Using Dual Tags with Tetracysteine Motifs, in Micro and Nano

Technologies in Bioanalysis. 2009. p. 421.

85. Lichty, J.J., et al., Comparison of affnity tags for protein purifcation. Protein

Expression and Purification, 2005. 41: p. 98-105.

86. Somers, E., J. Vanderleyden, and M. Srinivasan, Rhizosphere Bacterial Signaling:

A Love Parade Beneath Our Feet. Critical Reviews in Microbiology, 2004. 30: p.

205-240.

87. Borisov, I.V., et al., Changes in Azospirillum brasilense motility and the effect of

wheat seedling exudates. Microbiological Research, 2007.

88. Burdman, S., et al., Extracellular polysaccharide composition of Azospirillum

brasilense and its relation with cell aggregation. FEMS Microbiology Letters,

2000. 189(2): p. 259.

89. Baldani, V.L.D., et al., Establishment of inoculated Azospirillum spp. in the

rhizosphere and in roots of field grown wheat and sorghum. Plant Soil, 1986. 90:

p. 37-40.

90. Alexandre, G., S.E. Greer, and I.B. Zhulin, Energy Taxis is the dominant behavior

in Azospirillum brasilense. Journal of Bacteriology, 2000. 182(21): p. 6042-6048.

91. Steenhoudt, O. and J. Vanderleyden, Azospirillum, a free-living nitrogen-fixing

bacterium closely associated with grasses: genetic, biochemical and ecological

aspects. FEMS Microbiology Reviews, 2000. 24(4): p. 487-506.

92. Kamnev, A.A., et al., Effects of heavy metals on plant-associated rhizobacteria:

Comparison of endophytic and non-endophytic strains of Azospirillum brasilense.

Journal of Trace Elements in Medicine and Biology, 2005. 19(1): p. 91.

93. Assmus, B., et al., In Situ Localization of Azospirillum brasilense in the

Rhizosphere of Wheat with Fluorescently Labeled, rRNA-Targeted

Oligonucleotide Probes and Scanning Confocal Laser Microscopy. Appl.

Environ. Microbiol., 1995. 61: p. 1013-1019.

94. Tarrand, J.J., N.R. Krieg, and J. Do¨bereiner., A taxonomic study of the Spirillum

lipoferum group, with descriptions of a new genus, Azospirillum gen. nov., and

two species, Azospirillum lipoferum (Beijerinck) comb. nov. and Azospirillum

brasilense sp.nov. Can. J. Microbiol., 1978. 24: p. 967-980.

237

95. Martinez-Argudo, I., et al., Nitrogen fixation: key genetic regulatory mechanisms.

Biochem. Soc. Trans., 2005. 33(Pt 1): p. 152-156.

96. Leigh, J.A. and J.A. Dodsworth, Nitrogen Regulation in Bacteria and Archaea.

Annual Review of Microbiology, 2007. 61(1): p. 349-377.

97. Klipp, W., et al., eds. Genetics and Regulation of Nitrogen Fixation in Free-

Living Bacteria. 2004, Kluwer Academic Publishers.

98. Elmerich, C., et al., Regulation of NIF Gene Expression and Nitrogen Metabolism

in Azospirillum. Soil Biology and Biochemistry, 1997. 29(516): p. 847-852.

99. Forchhammer, K., Glutamine Signalling in Bacteria. Frontiers in Bioscience,

2007. 12: p. 358-370.

100. Zhulin, I.B., et al., Oxygen Taxis and Proton Motive Force in Azospirillum

brasilense. Journal of Bacteriology, 1996. 178(17): p. 5199-5204.

101. Srivastava, A. and A.K. Tripathi, Adenosine Diphosphate Ribosylation of

Dinitrogenase Reductase and Adenylylation of Glutamine Synthetase Control

Ammonia Excretion in Ethylenediamine-Resistant Mutants of Azospirillum

brasilense Sp7. Current Microbiology, 2006. 53: p. 317–323.

102. Chakraborty, B. and K.R. Samaddar, Evidence for the occurrence of an

alternative nitrogenase system in Azospirillum brasilense. FEMS Microbiology

Letters, 1995. 127: p. 127-131.

103. Eady, R.R. and R.L. Robson, Characteristics of N2 fixation in Mo-limited batch

and continuous cultures of Azotobacter vinelandii. Biochemisty Journal, 1984.

224: p. 853-862.

104. Fallik, E., Y.-K. Chan, and R.L. Robson, Detection of Alternative Nitrogenases in

Aerobic Gram-Negative Nitrogen-Fixing Bacteria. Journal of Bacteriology, 1991.

Vol. 173(1): p. 365-371.

105. Oda, Y., et al., Functional Genomic Analysis of Three Nitrogenase Isozymes in

the Photosynthetic Bacterium Rhodopseudomonas palustris. Journal of

Bacteriology, 2005. 187(22): p. 7784-7794.

106. Wadhams, G.H. and J.P. Armitage, Making Sense of it all: Bacterial Chemotaxis.

Nature Reviews Molecular Cell Biology, 2004. 5: p. 1024-1037.

107. Baker, M.D., P.M. Wolanin, and J.B. Stock, Systems biology of bacterial

chemotaxis. Current opinion in Microbiology, 2006. 9: p. 187-192.

108. Bren, A. and M. Eisenbach, How Signals Are Heard during Bacterial

Chemotaxis: Protein-Protein Interactions in Sensory Signal Propagation. Journal

of Bacteriology, 2000. 182(24): p. 6865-6873.

109. Baker, M.D., P.M. Wolanin, and J.B. Stock, Signal transduction in bacterial

chemotaxis. BioEssays, 2006. 28(1): p. 9-22.

110. Sourjik, V., Receptor clustering and signal processing in E. coli chemotaxis.

Trends in Microbiology, 2004. 12(12): p. 569-576.

111. Porter, S.L., G.H. Wadhams, and J.P. Armitage, Rhodobacter sphaeroides:

complexity in chemotactic signalling. Trends in Microbiology, 2008. 16(6): p.

251.

112. Larimer, F.W., et al., Complete genome sequence of the metabolically versatile

photosynthetic bacterium Rhodopseudomonas palustris. Nature Biotechnology,

2004. 22(1): p. 55.

238

113. Ferrandez, A., et al., Cluster II che Genes from Pseudomonas aeruginosa Are

Required for an Optimal Chemotactic Response. J. Bacteriol., 2002. 184(16): p.

4374-4383.

114. Hickman, J.W., D.F. Tifrea, and C.S. Harwood, A chemosensory system that

regulates biofilm formation through modulation of cyclic diguanylate levels.

Proceedings of the National Academy of Sciences of the United States of

America, 2005. 102(40): p. 14422-14427.

115. Galperin, M.Y., Structural Classification of Bacterial Response Regulators:

Diversity of Output Domains and Domain Combinations. Journal of Bacteriology,

2006. 188: p. 4169-4182.

116. Szurmant, H. and G.W. Ordal, Diversity in Chemotaxis Mechanisms among the

Bacteria and Archaea. Microbiol. Mol. Biol. Rev., 2004. 68(2): p. 301-319.

117. Bible, A.N., et al., Function of a Chemotaxis-Like Signal Transduction Pathway

in Modulating Motility, Cell Clumping, and Cell Length in the

Alphaproteobacterium Azospirillum brasilense. J. Bacteriol., 2008. 190(19): p.

6365-6375.

118. Stephens, B.B., S.N. Loar, and G. Alexandre, Role of CheB and CheR in the

Complex Chemotactic and Aerotactic Pathway of Azospirillum brasilense. J.

Bacteriol., 2006. 188(13): p. 4759-4768.

119. Elias, J.E., et al., Comparative evaluation of mass spectrometry platforms used in

large-scale proteomics investigations. Nature Methods, 2005. 2(9): p. 667-675.

120. Gingras, A.-C., et al., Analysis of protein complexes using mass spectrometry. Nat

Rev Mol Cell Biol, 2007. 8(8): p. 645.

121. Narasimhan, C., et al., MASPIC: Intensity-Based Tandem Mass Spectrometry

Scoring Scheme That Improves Peptide Identification at High Confidence.

Analytical Chemistry, 2005. 77(23): p. 7581.

122. Tanner, S., et al., InsPecT: Identification of Posttranslationally Modified Peptides

from Tandem Mass Spectra. Analytical Chemistry, 2005. 77(14): p. 4626.

123. Winnik, W.M. and P.A. Ortiz, Proteomic analysis optimization: Selective protein

sample on-column retention in reverse-phase liquid chromatographystar, open.

Journal of Chromatography B, 2008. 875(2): p. 478-486.

124. Busch, K.L., Space Charge in Mass Spectrometry. Spectroscopy, 2004. 19(6): p.

35-38.

125. Washburn, M.P., R.R. Ulaszek, and I. John R Yates, Reproducibility of

Quantitative Proteomic Analyses of Complex Biological Mixtures by

Multidimensional Protein Identification Technology. Analytical Chemistry, 2003.

75(19): p. 5054-5061.

126. Tabb, D.L., et al., MS2Grouper: Group Assessment and Synthetic Replacement of

Duplicate Proteomic Tandem Mass Spectra. Journal of the American Society for

Mass Spectrometry, 2005. 16(8): p. 1250.

127. Eng, J., Receiver Operating Characteristic Analysis: A Primer. Academic

Radiology, 2005. 12(7).

128. Boyer, M., et al., Bacteriophage Prevalence in the Genus Azospirillum and

Analysis of the First Genome Sequence of an Azospirillum brasilense Integrative

Phage. Appl. Environ. Microbiol., 2008. 74(3): p. 861-874.

239

129. Caballero-Melladoa, J., L. López-Reyes, and R. Bustillos-Cristales, Presence of

16S rRNA genes in multiple replicons in Azospirillum brasilense. FEMS

Microbiology Letters, 1999. 178 (2): p. 283-288.

130. Martin-Didonet, C.C.G., et al., Genome Structure of the Genus Azospirillum. J.

Bacteriol., 2000. 182(14): p. 4113-4116.

131. Katsy, E.I. and A.G. Prilipov, Mobile elements of an Azospirillum brasilense

Sp245 85-MDa plasmid involved in replicon fusions. Plasmid, 2009.

132. Onyeocha, I., et al., Physical map and properties of a 90-MDa plasmid of

Azospirillum brasilense Sp7. Plasmid, 1990. 23(3): p. 169.

133. Vanbleu, E., et al., Annotation of the pRhico plasmid of Azospirillum brasilense

reveals its role in determining the outer surface composition. FEMS

Microbiology Letters, 2004. 232(2): p. 165.

134. Altschul, S.F., et al., Gapped BLAST and PSI-BLAST: a new generation of protein

database search programs. Nucl. Acids Res., 1997. 25(17): p. 3389-3402.

135. Frazzon, J. and I.S. Schrank, Sequencing and complementation analysis of the

nifUSV genes from Azospirillum brasilense. FEMS Microbiology Letters, 1998.

159(2): p. 151.

136. Edgren, T. and S. Nordlund, Two pathways of electron transport to nitrogenase in

Rhodospirillum rubrum: the major pathway is dependent on the fix gene products.

FEMS Microbiology Letters, 2006. 260(1): p. 30-35.

137. Sperotto, R.A., et al., The Electron Transfer Flavoprotein fixABCX Gene

Products from Azospirillum brasilense Show a NifA-Dependent Promoter

Regulation. Current Microbiology 2004. 49: p. 267-273.

138. Edgren, T. and S. Nordlund, The fixABCX Genes in Rhodospirillum rubrum

Encode a Putative Membrane Complex Participating in Electron Transfer to

Nitrogenase. J. Bacteriol., 2004. 186(7): p. 2052-2060.

139. Hauwaerts, D., et al., A major chemotaxis gene cluster in Azospirillum brasilense

and relationships between chemotaxis operons in alpha-proteobacteria. FEMS

Microbiology Letters, 2002. 208: p. 61-67.

140. Xie, Z., et al., PAS domain containing chemoreceptor couples dynamic changes in

metabolism with chemotaxis. Proceedings of the National Academy of Sciences.

107(5): p. 2235-2240.

141. White, D., The Physiology and Biochemistry of Prokaryotes. 3 ed. 2007: Oxford

University Press, Inc. 628.

142. Chen, G., H. Zhu, and Y. Zhang, Soil microbial activities and carbon and

nitrogen fixation. Research in Microbiology, 2003. 154(6): p. 393.

143. Fadel-Picheth, C.M.T., et al., Regulation of Azospirillum brasilense nifA gene

expression by ammonium and oxygen. FEMS Microbiology Letters, 1999. 179: p.

281-288.

144. Rehder, D., Vanadium nitrogenase. Journal of Inorganic Biochemistry, 2000.

80(1-2): p. 133.

145. Ignoul, S. and J. Eggermont, CBS domains: structure, function, and pathology in

human proteins. Am J Physiol Cell Physiol, 2005. 289(6): p. C1369-1378.

240

146. Mahmood, N.A.B.N., E. Biemans-Oldehinkel, and B. Poolman, Engineering of

ion sensing by the CBS module of the ABC transporter OpuA. J. Biol. Chem.,

2009: p. M901238200.

147. Ma, B.-G., et al., Characters of very ancient proteins. Biochemical and

Biophysical Research Communications, 2008. 366(3): p. 607.

148. I. Kefalogianni, G.A., Modeling growth and biochemical activities of

Azospirillum spp. Applied Microbiology and Biotechnology, 2002. 58(3): p. 352.

149. Tsagou, V. and G. Aggelis, Growth dynamics of Azospirillum lipoferum at steady

and transitory states in the presence of NH. Journal of Applied Microbiology,

2006. 100(2): p. 286-295.

150. Antoine, R., et al., Overrepresentation of a Gene Family Encoding

Extracytoplasmic Solute Receptors in Bordetella. J. Bacteriol., 2003. 185(4): p.

1470-1474.

151. Marchler-Bauer, A., et al., CDD: specific functional annotation with the

Conserved Domain Database. Nucleic Acids Research, 2009(37(Database issue)):

p. D205-10.

152. Dunwell, J.M., S. Khuri, and P.J. Gane, Microbial Relatives of the Seed Storage

Proteins of Higher Plants: Conservation of Structure and Diversification of

Function during Evolution of the Cupin Superfamily. Microbiol. Mol. Biol. Rev.,

2000. 64(1): p. 153-179.

153. Ona, O., et al., Growth and indole-3-acetic acid biosynthesis of Azospirillum

brasilense Sp245 is environmentally controlled. FEMS Microbiology Letters,

2005. 246(1): p. 125.

154. Kamnev, A.A., et al., Responses of Azospirillum brasilense to Nitrogen

Deficiency and to Wheat Lectin: A Diffuse Reflectance Infrared Fourier

Transform (DRIFT) Spectroscopic Study. Microbial Ecology, 2008. 56: p. 615-

624.

155. Freestone, P., et al., The universal stress protein, UspA, of Escherichia coli is

phosphorylated in response to stasis. Journal of Molecular Biology, 1997. 274(3):

p. 318.

156. Nachin, L., U. Nannmark, and T. Nystro¨m, Differential Roles of the Universal

Stress Proteins of Escherichia coli in Oxidative Stress Resistance, Adhesion, and

Motility. Journal of Bacteriology, 2005. 187(18): p. 6265–6272.

157. Blaha, G., Carlos, and I. Schrank, An Azospirillum brasilense Tn5 mutant with

modified stress response and impaired in flocculation. Antonie van

Leeuwenhoek, 2003. 83(1): p. 35.

158. Kefalogianni, I. and G. Aggelis, Modeling growth and biochemical activities of

Azospirillum spp. Applied Microbiology and Biotechnology, 2002. 58(3): p. 352.

159. Crans, D.C., et al., The Chemistry and Biochemistry of Vanadium and the

Biological Activities Exerted by Vanadium Compounds. Chem. Rev., 2004.

104(2): p. 849-902.

160. Rehder, D., Structure and function of vanadium compounds in living organisms.

BioMetals, 1992. 5: p. 3-12.

241

161. Guerrieri, N., et al., Vanadium inhibition of serine and cysteine proteases.

Comparative Biochemistry and Physiology - Part A: Molecular & Integrative

Physiology, 1999. 122(3): p. 331.

162. Fisher, K., D.J. Lowe, and J. Petersen, Vanadium(V) is reduced by the ‘as

isolated’ nitrogenase Fe-protein at neutral pH. Chemical Communications, 2006:

p. 2807-2809.

163. Lyalikova, N.N. and N.A. Yurkova, Role of Microorganisms in Vanadium

Concentration and Dispersion. Geomicrobiology Journal, 1992. 10: p. 15-26.

164. Bellenger, J.P., T. Wichard, and A.M.L. Kraepiel, Vanadium Requirements and

Uptake Kinetics in the Dinitrogen-Fixing Bacterium Azotobacter vinelandii. Appl.

Environ. Microbiol., 2008. 74(5): p. 1478-1484.

165. Wilkinson, S.P. and A. Grove, Ligand-responsive Transcriptional Regulation by

Members of the MarR Family of Winged Helix Proteins. Current Issues in

Molecular Biology, 2006. 8: p. 51-62.

166. González, P.J., et al., Bacterial nitrate reductases: Molecular and biological

aspects of nitrate reduction. Journal of Inorganic Biochemistry, 2006. 100(5-6): p.

1015.

167. Taté, R., et al., Cloning and transcriptional analysis of the lipA (lipoic acid

synthetase) gene from Rhizobium etli. FEMS Microbiology Letters, 1997. 149(2):

p. 165.

168. Martin, N., et al., A Mutant in lipA (yutB), encoding Lipoic Acid Synthase,

Provides Insight into the Interplay Between Branched-Chain and Unsaturated

Fatty Acids Biosynthesis in Bacillus subtilis. J. Bacteriol., 2009: p. JB.01160-09.

169. Galimand, M., et al., Identification of DNA regions homologous to nitrogen

fixation genes nifE, nifUS and fixABC in Azospirillum brasilense Sp7. Journal of

General Microbiology, 1989. 135(5): p. 1047-1059.

170. Maddocks, S.E. and P.C.F. Oyston, Structure and function of the LysR-type

transcriptional regulator (LTTR) family proteins. Microbiology, 2008. 154(12): p.

3609-3623.

171. Lerner, A., et al., The Azospirillum brasilense Sp7 noeJ and noeL genes are

involved in extracellular polysaccharide biosynthesis. Microbiology, 2009: p.

mic.0.031807-0.

172. Lerner, A., Y. Okon, and S. Burdman, The wzm gene located on the pRhico

plasmid of Azospirillum brasilense Sp7 is involved in lipopolysaccharide

synthesis. Microbiology, 2009. 155(3): p. 791-804.

173. Bahat-Samet, E., S. Castro-Sowinski, and Y. Okon, Arabinose content of

extracellular polysaccharide plays a role in cell aggregation of Azospirillum

brasilense. FEMS Microbiology Letters, 2004. 237(2): p. 195.

174. Bos, M.P., V. Robert, and J. Tommassen, Biogenesis of the Gram-Negative

Bacterial Outer Membrane. Annual Review of Microbiology, 2007. 61(1): p. 191-

214.

175. Ogura, M., et al., A new Bacillus subtilis gene, med, encodes a positive regulator

of comK. J. Bacteriol., 1997. 179(20): p. 6244-6253.

176. Deka, R.K., et al., The PnrA (Tp0319; TmpC) Lipoprotein Represents a New

Family of Bacterial Purine Nucleoside Receptor Encoded within an ATP-binding

242

Cassette (ABC)-like Operon in Treponema pallidum. J. Biol. Chem., 2006.

281(12): p. 8072-8081.

177. McBroom, A.J., et al., Outer Membrane Vesicle Production by Escherichia coli Is

Independent of Membrane Instability. J. Bacteriol., 2006. 188(15): p. 5385-5392.

178. Yu, F., S. Inouye, and I. M, Lipoprotein-28, a cytoplasmic membrane lipoprotein

from Escherichia coli. Cloning, DNA sequence, and expression of its gene.

Journal of Biological Chemistry, 1986. 261: p. 2284-2288.

179. Rao, D.E.C.S., et al., Molecular characterization, physicochemical properties,

known and potential applications of phytases: An overview. Critical Reviews in

Biotechnology, 2009. 29(2): p. 182 - 198.

180. Lim, B.L., et al., Distribution and diversity of phytate-mineralizing bacteria.

ISME J, 2007. 1(4): p. 321.

181. Paoletti, A.C., et al., Quantitative proteomic analysis of distinct mammalian

Mediator complexes using normalized spectral abundance factors. Proceedings of

the National Academy of Sciences, 2006. 103(50): p. 18928-18933.

182. Lamarche, M.G., et al., The phosphate regulon and bacterial virulence: a

regulatory network connecting phosphate homeostasis and pathogenesis. FEMS

Microbiol Rev, 2008. 32(3): p. 461 - 473.

183. Sperandeo, P., G. Dehò, and A. Polissi, The lipopolysaccharide transport system

of Gram-negative bacteria. Biochimica et Biophysica Acta (BBA) - Molecular

and Cell Biology of Lipids, 2009. 1791(7): p. 594.

184. Doyle, S.M. and S. Wickner, Hsp104 and ClpB: protein disaggregating

machines. Trends in Biochemical Sciences, 2009. 34(1): p. 40.

185. Wood, J., Osmosensing by bateria. Science STKE, 2006. 357: p. pe43.

186. Grage, K., et al., Bacterial Polyhydroxyalkanoate Granules: Biogenesis,

Structure, and Potential Use as Nano-/Micro-Beads in Biotechnological and

Biomedical Applications. Biomacromolecules, 2009. 10(4): p. 660.

187. Potter, M. and A. Steinbuchel, Poly(3-hydroxybutyrate) Granule-Associated

Proteins: Impacts on Poly(3-hydroxybutyrate) Synthesis and Degradationâ€.

Biomacromolecules, 2005. 6(2): p. 552.

188. VanBogelen, R.A., et al., Global analysis of proteins synthesized during

phosphorus restriction in Escherichia coli. J. Bacteriol., 1996. 178(15): p. 4344-

4366.

189. Mendel, R.R., Biology of the molybdenum cofactor. J. Exp. Bot., 2007. 58(9): p.

2289-2296.

190. Kisker, C., a.H. Schindelin, and D.C. Rees, Molybdenum-Cofactor containing

Enzymes:Structure and Mechanism. Annual Review of Biochemistry, 1997.

66(1): p. 233-267.

191. Raux, E., H.L. Schubert, and M.J. Warren , Biosynthesis of cobalamin (vitamin

B12): a bacterial conundrum. Cellular and Molecular Life Sciences, 2000.

57(13): p. 1880.

192. Taylor, B.L. and I.B. Zhulin, PAS Domains: Internal Sensors of Oxygen, Redox

Potential, and Light. Microbiol. Mol. Biol. Rev., 1999. 63(2): p. 479-506.

193. Galperin, M.Y., Bacterial signal transduction network in a genomic perspective.

Environmental Microbiology, 2004. 6(6): p. 552-567.

243

194. Scheffers, D.-J. and M.G. Pinho, Bacterial Cell Wall Synthesis: New Insights

from Localization Studies. Microbiology and Molecular Biology Reviews, 2005.

Vol. 69 (No. 4): p. 585–607.

195. Blackburn, N.T. and A.J. Clarke, Identification of Four Families of Peptidoglycan

Lytic Transglycosylases. Journal of Molecular Evolution, 2001. 52(1): p. 78.

196. Garcıa-Calderon, C.B., J. Casadesus, and F. Ramos-Morales, Rcs and PhoPQ

Regulatory Overlap in the Control of Salmonella enterica Virulence. Journal of

Bacteriology, 2007. 189(18): p. 6635–6644.

197. Kumar, A.S., K. Mody, and B. Jha, Bacterial exopolysaccharides - a perception.

Journal of Basic Microbiology, 2007. 47(2): p. 103-117.

198. Wu, R., et al., Crystal Structure of Bacillus anthracis Transpeptidase Enzyme

CapD. Journal of Biological Chemistry, 2009. 284(36): p. 24406-24414.

199. Candela, T. and A. Fouet, Bacillus anthracis CapD, belonging to the gamma-

glutamyltranspeptidase family, is required for the covalent anchoring of capsule

to peptidoglycan. Molecular Microbiology, 2005. 57(3): p. 717-726.

200. Santhanagopalan, V., C. Coker, and S. Radulovic, Characterization of RP 333, a

gene encoding CapD of Rickettsia prowazekii with UDP-glucose 4-epimerase

activity. Gene, 2006. 369: p. 119-125.

201. Lerner, A., et al., Glycogen phosphorylase is involved in stress endurance and

biofilm formation in Azospirillum brasilense Sp7. FEMS Microbiology Letters,

2009. 300(1): p. 75-82.

202. Burdman, S., Y. Okon, and E. Jurkevitch, Surface Characteristics of Azospirillum

brasilense in Relation to Cell Aggregation and Attachment to Plant Roots. Critical

Reviews in Microbiology, 2000. 26(2): p. 91 - 110.

203. Filloux, A., A. Hachani, and S. Bleves, The bacterial type VI secretion machine:

yet another player for protein transport across membranes. Microbiology, 2008.

154(6): p. 1570-1583.

204. Pukatzki, S., S.B. McAuley, and S.T. Miyata, The type VI secretion system:

translocation of effectors and effector-domains. Current Opinion in Microbiology,

2009. 12(1): p. 11.

205. Roest, H.P., et al., A Rhizobium leguminosarum Biovar trifolii Locus Not

Localized on the Sym Plasmid Hinders Effective Nodulation on Plants of the Pea

Cross-Inoculation Group. Molecular Plant-Microbe Interactions, 1997. 10(7): p.

938-941.

206. Bingle, L.E.H., C.M. Bailey, and M.J. Pallen, Type VI secretion: a beginner's

guide. Current Opinion in Microbiology, 2008. 11(1): p. 3.

207. Cascales, E., The type VI secretion toolkit. EMBO reports, 2008. 9(8): p. 735-741.

208. Leiman, P.G., et al., Type VI secretion apparatus and phage tail-associated

protein complexes share a common evolutionary origin. Proceedings of the

National Academy of Sciences, 2009. 106(11): p. 4154-4159.

209. Filloux, A., The type VI secretion system: a tubular story. EMBO J, 2009. 28(4):

p. 309.

210. Weber, B., et al., Type VI secretion modulates quorum sensing and stress

response in Vibrio anguillarum. Environmental Microbiology, 2009. 11(12): p.

3018-3028.

244

211. Syed, K.A., et al., The Vibrio cholerae Flagellar Regulatory Hierarchy Controls

Expression of Virulence Factors. J. Bacteriol., 2009. 191(21): p. 6555-6570.

212. Zhou, L., J. Wang, and L.-H. Zhang, Modulation of Bacterial Type III Secretion

System by a Spermidine Transporter Dependent Signaling Pathway. PLoS One,

2007. 12: p. e1291.

213. Kadouri, D., E. Jurkevitch, and Y. Okon, Involvement of the Reserve Material

Poly-{beta}-Hydroxybutyrate in Azospirillum brasilense Stress Endurance and

Root Colonization. Appl. Environ. Microbiol., 2003. 69(6): p. 3244-3250.

214. Busso, D., B. Delagoutte-Busso, and D. Moras, Construction of a set Gateway-

based destination vectors for high-throughput cloning and expression screening

in Escherichia coli. Analytical Biochemistry, 2005. 343(2): p. 313.

215. Melcher, K., A Modular Set of Prokaryotic and Eukaryotic Expression Vectors.

Analytical Biochemistry, 2000. 277(1): p. 109.

216. Freuler, F., et al., Development of a novel Gateway-based vector system for

efficient, multiparallel protein expression in Escherichia coli. Protein Expression

and Purification. 2008 59(2):p.232

217. Rocco, C.J., et al., Construction and use of new cloning vectors for the rapid

isolation of recombinant proteins from Escherichia coli. Plasmid, 2008. 59(3): p.

231.

218. Nagarajan, V., et al., Modular expression and secretion vectors for Bacillus

subtilis. Gene, 1992. 114(1): p. 121.

219. Krogan, N.J., et al., Global landscape of protein complexes in the yeast

Saccharomyces cerevisiae. Nature, 2006. 440(7084): p. 637.

220. Séraphin, B., et al., Tandem Affinity Purification to Enhance Interacting Protein

Identification, in Protein–Protein Interactions: A Molecular Cloning Manual,

Chapter 17. 2002, Cold Spring Harbor Laboratory Press. p. 313-328.

221. Fodor, B.D., et al., Modular Broad-Host-Range Expression Vectors for Single-

Protein and Protein Complex Purification. Applied and Environmental

Microbiology, 2004. 70(2): p. 712-721.

222. Antoine, R. and C. Locht, Isolation and molecular characterization of a novel

broad-host-range plasmid from Bordetella bronchiseptica with sequence

similarities to plasmids from gram-positive organisms. Molecular Microbiology,

1992. 6(13): p. 1785-1799.

223. Kovach, M.E., et al., Four new derivatives of the broad-host-range cloning vector

pBBR1MCS, carrying different antibiotic-resistance cassettes. Gene, 1995.

166(1): p. 175.

224. Lynch, M.D. and R.T. Gill, Broad host range vectors for stable genomic library

construction. Biotechnology and Bioengineering, 2006. 94(1): p. 151-158.

225. Hartley, J.L., G.F. Temple, and M.A. Brasch, DNA Cloning Using In Vitro Site-

Specific Recombination. Genome Res., 2000. 10(11): p. 1788-1795.

226. Quandt, J. and M.F. Hynes, Versatile suicide vectors which allow direct selection

for gene replacement in Gram-negative bacteria. Gene, 1993. 127: p. 15-21.

227. Cozzarelli, N.R., R.B. Kelly, and A. Kornberg, A Minute Circular DNA from

Escherichia Coli 105*. BIochemistry, 1968. 60: p. 992-999.

245

228. Simon, R., U. Priefer, and A. Puhler, A Broad Host Range Mobilization System

for In Vivo Genetic Engineering: Transposon Mutagenesis in Gram Negative

Bacteria. Nat Biotech, 1983. 1(9): p. 784.

229. Sambrook, J. and D.W. Russell, Molecular cloning: a laboratory manual. 2001,

Cold Spring Harbor, N.Y.: Cold Spring Harbor Laboratory Press.

230. Schmidt, T.G.M., et al., Molecular Interaction Between the Strep-tag Affinity

Peptide and its Cognate Target, Streptavidin. Journal of Molecular Biology,

1996. 255(5): p. 753-766.

231. Dougherty, W.G., S.M. Cary, and T.D. Parks, Molecular Genetic Analysis of a

Plant Virus Polyprotein Cleavage Site: A Model. Virology, 1989. 171: p. 356-

364.

232. Vieira, J. and J. Messing, The pUC plasmids, an M13mp7-derived system for

insertion mutagenesis and sequencing with synthetic universal primers. Gene,

1982. 19(3): p. 259.

233. Giannone, R.J., et al., Dual-tagging system for the affinity purification of

mammalian protein complexes. Biotechniques, 2007. 43(3)(3): p. 296-300.

234. Berrow, N.S., et al., A versatile ligation-independent cloning method suitable for

high-throughput expression screening applications. Nucl. Acids Res., 2007.

35(6): p. e45-.

235. Cabrita, L.D., W. Dai, and S.P. Bottomley, A family of E. coli expression vectors

for laboratory scale and high throughput soluble protein production. BMC

Biotechnology, 2006. 6(12).

236. Chanda, P.K., W.A. Edris, and J.D. Kennedy, A set of ligation-independent

expression vectors for co-expression of proteins in Escherichia coli. Protein

Expression and Purification, 2006. 47(1): p. 217.

237. Donnelly, M.I., et al., An expression vector tailored for large-scale, high-

throughput purification of recombinant proteins. Protein Expression and

Purification, 2006. 47(2): p. 446.

238. Stols, L., et al., A New Vector for High-Throughput, Ligation-Independent

Cloning Encoding a Tobacco Etch Virus Protease Cleavage Site. Protein

Expression and Purification, 2002. 25(1): p. 8.

239. Braun, P., et al., Proteome-scale purification of human proteins from bacteria.

Proceedings of the National Academy of Sciences, 2002. 99(5): p. 2654-2659.

240. Abu-Farha, M., F. Elisma, and D. Figeys, Identification of Protein–Protein

Interactions by Mass Spectrometry Coupled Techniques. 2008.

241. Waugh, D.S., Making the most of affinity tags. Trends in Biotechnology, 2005.

23(6): p. 316.

242. Arnau, J., et al., Current strategies for the use of affinity tags and tag removal for

the purification of recombinant proteins. Protein Expression and Purification,

2006. 48(1): p. 1.

243. Gloeckner, C.J., et al., A novel tandem affinity purification strategy for the

efficient isolation and characterisation of native protein complexes. Proteomics,

2007. 7(23): p. 4228-4234.

246

244. Parks, T.D., et al., Release of Proteins and Peptides from Fusion Proteins Using a

Recombinant Plant Virus Proteinase. Analytical Biochemistry, 1994. 216(2): p.

413.

245. Jiao, Y. and D.K. Newman, The pio Operon Is Essential for Phototrophic Fe(II)

Oxidation in Rhodopseudomonas palustris TIE-1. J. Bacteriol., 2007. 189(5): p.

1765-1773.

246. Braatsch, S., et al., Rhodopseudomonas palustris CGA009 Has Two Functional

ppsR Genes, Each of Which Encodes a Repressor of Photosynthesis Gene

Expression. Biochemistry, 2006. 45(48): p. 14441.

247. Alberts, B., et al., Molecular Biology of the Cell. 4 ed. 2002: Garland Science.

1463.

248. Savidor, A., et al., Expressed Peptide Tags: An Additional Layer of Data for

Genome Annotation. Journal of Proteome Research, 2006. 5(11): p. 3048.

247

Vita

Gurusahai Khalsa-Moyers grew up in Clinton, Tennessee, and still resides in East

Tennessee. She attended the University of Tennessee where she obtained her degree in

Electrical Engineering. When her children arrived, she elected to stay home with them,

homeschooling them until they reached the eighth grade. When her youngest child

entered eighth grade, she returned to the University of Tennessee to further her education

in a subject that had always been of interest to her, biochemistry. Because of the 16-year

time period away from an academic setting, it was necessary for her to take some

background courses before beginning graduate school. In January 2004, Gurusahai

entered the Genome science and Technology program at the University of Tennessee.

During the process of obtaining her PhD, she re-discovered her initial love for

technology, and became fascinated with the application of analytical technology to

microbial systems.